Monitoring Narragansett Bay Oil Spills by Infrared Spectroscopy Chris W. Brown,* Patricia F. Lynch, and Mark Ahmadjian Department of Chemistry, University of Rhode Island, Kingston, R.I. 02881
A 1973 oil spill on Narragansett Bay was monitored for 30 days using infrared spectroscopy. All samples of the spill collected during this period were traced to the source of the spill by computer matching of their infrared spectra. Two subsequent oil slicks found in the same general area were shown to be from a different source by the same method of analysis.
During the evening of April 9, 1973, an oil tanker struck an underwater object in Narragansett Bay approximately 10 miles from its destination in East Providence, R.I. The tanker was extensively damaged and -100,000 gal. of No. 6 grade fuel oil were released into the Bay. On April 11, less than 48 hr after the spill, we started sampling the oil from the beaches and shallow waters near the shore. In this report we show that the source of the oil can be identified by infrared spectroscopy after more than 30 days of weathering on the open sea. The infrared bands used to identify the oil and the computer analysis of the spectra have been discussed previously (1). Complete infrared spectra from 4000-400 cm-l are recorded; however, 21 selected bands between 1200 and 650 cm--l are used for the identification. For each of the known samples the peak heights of the 21 bands are converted into absorptivities, and these are stored in a computer file (if a band is not observed it is given a zero absorptivity and is eliminated from the analysis). To match an unknown with the correct known, the following procedure is carried out: The absorptivities of the bands in the spectrum of the unknown are calculated, the ratios of the absorptivities for an unknown to those of
each known are calculated, these ratios are averaged for each known stored in the computer file, the percent difference between this average and each ratio is determined, and finally the number of bands with a percent difference less than a certain value (e.g., 5, 10, 25%) are listed. The unknown is matched to a known when most of its bands are within 5 or 10%. Other methods to identify the source of oil spills by infrared spectroscopy have been suggested (2-9). The present analysis differs from the others in the type and number of bands used, and in the method of computer analysis. By use of our method, an unknown can be matched to the correct known (assuming that the required data for the known are stored in the computer file) in approximately 30 min; this includes the time required to measure the spectrum and perform computer analysis of the data.
Experimental Infrared spectra of the samples taken from the tanker, the beaches, and the water were measured on a PerkinElmer Model 521 infrared spectrometer. The spectral slit width of the instrument was
