sidering the retention volume for the specific chemical class, our technique is obviously more useful for more complex mixtures. This is particularly so for high boiling compounds, for which little information regarding retention values is reported. Walsh and Merritt have shown that, in those cases in which it is possible to make use of the correlation plots like log retention volume 21s. carbon number for homologous series, the knowledge of the component
functionality can lead to its exact identification. As an additional feature, the device described here allows in some cases the control of the purity of a gas chromatographic peak. When a light-colored spot corresponds to a large peak, this may be due to the peak being the result of more than one component. I n this case one should also t r y other reagents. However, it may be that the light color of the spot is due to a single compound
in which the functional group is diluted in the molecule. I n making this evaluation one must consider the effect of the nonlinear response of a gas chromatographic detector for compounds of different classes and for different molecular weight members of a n homologous series. ACKNOWLEDGMENT
The authors thank Mario Psallidi for technical help.
An Improved Digital Readout Method Ralph Eno, Applied Physics Corp., Monrovia, Calif.
people, such as R . E. Biggers and G. P. Smith of Oak Ridge National Laboratory. J. E. Mapes of Brookhaven National Laboratories, and R. C. Molter of Libby-Owens Ford, are using digital readout equipment with spectrophotometers, gas chromatographs, and other types of instrumentation. Invariably, they use the same basic system of readout for the data (3) lvhich involves separating the wavelength or time axis into equal intervals. At each interval, a reading of the recorder pen deflection and wavelength is recorded. When using this method with a spectrophotometer, i t is necessary that readout be made a t intervals not greater than the instrument resolution to avoid loss of spectral information. Therefore, the recording of an ultraviolet-visible spectra (1850A. to 655OA.) on a high resolution instrument, would require the recording of about 4700 points tb characterize the curve. With this method of recording. the sections of the curve without spectral information are read out as often as those of high spectral information. An extreme example of this is shown in Figure 1, which is the visible transmission curve of an interference filter. Using the above method of readout, 3500 poirits would have been recorded, but only 950 points T o d d have fallen in areas of spectral information. Reducing the number of points would reduce the time needed by a computer to identify a chemical compound by a like factor. An alternative method of recording data is proposed. This method produces data points in greatest number in those sections of the curve having the highest spectral information. Rather than using equal intervals of wavelength, the absorbance (or transmittance) scale is divided into equal intervals a t which the absorbance value and wavelength are recorded. I n the curve shown in the figure, had each interval been 0.25y0transmittance, ANY
15 16
ANALYTICAL CHEMISTRY
only 745 points n-ould have h e n recorded and all but four would have fallen in the section of spectral information. A further refinement of this method can be made. Nothing is gained by recording the points falling on the steep sides of peaks so long as the recorder pen motor is running a t mayimum speed. Under this condition, the pen merely draws a straight line. T o eliminate these points, a derivative sensing device is placed on the pen system that prevents it from recording points when the rate-of-change of absorbance (or transmittance) with respect to wavelength is greater than some given value. One of many such devices is an A-C tachometer ( 2 ) attached to the recorder pen drive motor. The rectified output from the tachometer is then used to adjust the readout rate of the system. Details about the system can best be obtained from the manufacturer of digital systems ( I , 4 ) . KOloss of spectral information occurs with this equipment, becausp the pen is moving a t less than full speed on the skirts, shoulders, and a t points of inflection of the curve. By adjusting the derivative sensing equipment so that it will allow readout, except in the shaded area shown in the figure, the number of data points is cut to about 220. The high initial cost of derivatiLe sensing equipment may discourage many prospective users, but fortunately there is an inexpensive compromise-a device
which limits the readout rate to some maximum value and permits recording data points a t even intervals of pen deflection. One of the slower IBM card punches has a maximum punching rate of 30 data points per second, thus allowing a maximum of 30 points during a full chart maximum velocity excursion of the recorder pen. This is an adequate punch rate because a t normal scanning speeds for spectrophotometers, it vr-ould allow readout a t intervals of one third of instrument resolution. Points of inflection on the steep slopes of curves would be characterized by a number of points having small differences in absorbance due t o the lower velocity of the pen in this area. This device xould note all the points recorded using the derivative sensing device, yet would allow only 30 extra point. to be recorded in the area of low spectral information. X o loss of precision or spectral information occurs with the compromise> method. At the peaks of curves, the pen will be moving s l o ~ l yand allonthe punch to read out a t each equal change in absorbance. Weak peaks will be seen if readout is made for equal intervals of absorbance, not larger than about two to three times the s!*stem noise. Precise records of absorbance or resolution can be made because data points in the sections of low spectral information are eliminated. The application of either of the improved methods of digital readout TT ill significantly reduce the amount of computer memory required to stow spectral data and will reduce computation time to a more realistic figure. LITERATURE CITED
Corp., Monrovia, Calif. itzgerald, A. E., Kingsley, C., w t r i c Machinery,” p. 468, McGrawr k , 1952. D., R.C.A. Engineer, 7 , No. 1, p. 49, June-July 1961. (4) Radio Corp. of America, Moorestown, N. J.
i l ’i Datex
Figure 1. Transmittance curve for Spectrolab Type 2347 Interference Filter
