indicates that this percentage is the maximum performance which could be expected. If the number of tubes were increased to 50, it is predicted that nearly 100% of the methyl linolenate would be recovered (Figure 4). From Figure 3 it can be seen that the maximum weight based on an extraction coefficient of 1.0 in a 50-tube apparatus would be 25 times the feed rate and for 25 tubes 13 times the feed rate, so that if equal concentrations were to be maintained and 50 tubes (99.9% recovery) were used, the feed rate would have to be about one half that used for 25 tubes (50 % recovery). Therefore, there would be no change in linolenate production rate because of the interaction between maximum feed and recovery. If 32 tubes were used, 90 % of the methyl linolenate in linseed oil methyl esters would be recovered, and the feed rate would have to be reduced to three fourths that used for 25 tubes. Thus, a production rate almost 1.4 times greater than that on a 25- or 50-tube machine is predicted. It is noteworthy that the most efficient recovery does not always give the maximum production rate.
The example of methyl linolenate production is simplified, and if one takes into account all components and their extraction coefficients in combination, predictions are fairly accurate. Also, it is sometimes profitable to overload the machine, if enough tubes are available, to correct for possible changes in extraction coefficients. Use of the relationship presented eliminates much of the trial and error previously associated with CDCD by permitting nearly optimal conditions to be tried first.
RECEIVED for review November 21,1966. Accepted January 30,1967. Presented before the Division of Analytical Chemistry at the 152nd National Meeting of the American Chemical Society, New York, N.Y., September 11-16, 1966. The Northern Laboratory is headquarters for the Northern Utilization Research and Development Division, Agricultural Research Service, U.S. Department of Agriculture.
Time-Resolved Phosphorimetry as a Method of Simultaneous Analysis of Two-Component Mixtures P. A. St. John and J. D. Winefordner Department of Chemistry, University of Florida, Gainesville, FIa. 32601 Time-resolved phosphorimetry is used to measure twocomponent mixtures of phosphorescent compounds. The method utilizes differences in decay times of the phosphors. The exponentially decaying phosphors are resolved by using a logarithmic responding instrument. The concentration of the two components may be determined from the recording of the logarithm of the phosphorescence signal due to the mixture vs. time after termination of excitation, by a method analogous to that used in radioactive isotopic analysis. Mixtures of tryptophan-tryosine and benzoic acid-benzaldehyde are measured using this technique with an overall relative recovery error of about 3%.
THEUSE of a semilogarithmic plot of disintegration rate us. time after termination of activation for isotopes differing in half lives is a simple means of measuring radioactive isotopic species and their concentrations ( I ) . This paper describes the application of this technique to the measurement of the concentrations of two-component mixtures of phosphorescent organic compounds differing in decay times. This technique in phosphorimetry will be called time-resolved phosphorimetry. Keirs, Britt, and Wentworth (2) described a technique called phosphoroscopic resolution which also utilized differences in phosphorescence decay times as a means of selective analysis. In their technique, compounds differing in decay times were resolved by means of variation of the phosphoroscope shutter delay time (3). Their method, however, was applicable only to compounds with fairly short decay times (in the millisecond range). Time-resolved phosphorimetry, on the other hand, is applicable to compounds having millisecond and longer decay times. (1) G. E. Boyd, ANAL.CHEM., 21, 335 (1949). (2) R. J. Keirs, R. D. Britt, and W. E. Wentworth, Ibid., 29, 202
(1957).
(3) T. C . O'Haver and J. D. Winefordner, Zbid., 38, 602 (1966).
500
ANALYTICAL CHEMISTRY
Time-resolved phosphorimetry will be especially useful for small concentrations of compounds which are difficult to separate physically, which are difficult to measure by most instrumental methods, and which have similar excitation and emission phosphorescence spectra. Two examples of timeresolved phosphorimetry are given here to illustrate the method. Mixtures of benzoic acid-benzaldehyde and mixtures of tryptophan-tyrosine in various concentration ratios were measured by this new technique.
THEORY The phosphorescence spectra of the majority of organic molecules is produced by radiative transitions from the first excited triplet electronic state to the ground singlet electronic state. Many phosphorescent molecules exhibit exponential decay when the exciting radiation is terminated (4). The phosphorescence intensity, I , as a function of time, t , following termination of excitation of a multicomponent mixture of independently decaying phosphors (organic or inorganic) is given (4) by
z = X I ( = Z I Pexp(-t/rt)
(1)
where I: is the intensity of phosphor, i, at time t = 0, and r t is the decay time of phosphor, i. For a two-component system (species A and B), Z is given by
I = IA
+ le = IA'
exp(-t/rA)
+ IB'
exp(-t/rB)
(2)
The phosphorescence intensity of a single component is directly related to the excitation intensity a: described by St. John, McCarthy, and Winefordner (5). However, in an (4) T. Forster, Fluoresrenz Organischen Verbindungen, Vanden-
hoeck, Ruprecht, 1951.
(5) P. A. St. John, W. J. McCarthy, and J. D. Winefordner, ANAL. CHEM., 38, 1828 (1966).
X-Y Recorder
1P21 PM.TUBE
Figure 1. Block diagram of instrumentation used for timeresolved phosphorimetry
actual experimental measurement of a multicomponent system, the intensity of exciting radiation for species, i, is reduced by absorption (of the exciting radiation by the other species present. This correction will be described below. EXPERIMENTAL
Apparatus. An Aminco-Bowman spectrophotofluorometer (American Instrument Co., Inc., Silver Spring, Md.) with a phosphoroscope attachment was used to obtain the phosphorescence. The monochromator slit program was 5 mm, 4 mm, 4 mm, 5 mm, 5 mm, as specified by the manufacturer (6). The excitation source was a 150-watt xenon arc lamp. A manual guillotine shutter was mounted at the first excitation slit (first 5-mm slit) to terminate the exciting radiation. A block diagram of the electronic system used to record the logarithm of the phosphorescence signal as a function of time is given in Figure 1. An Aminco photomultiplier photometer and a poiled RCA 1P21 multiplier phototube were used to detect the phosphorescence radiation. The 50 mv output of the photometer was fed to a low pass filter40 Hz chopper network. The a.c. output was amplified 100fold by a preamplifier (Model 122, Tektronix, Inc., Portland, Ore.) set for narrow bandwidth, and then passed through a 120 Hz twin-T filter. The filtered a.c. signal was introduced into a logarithmic converter (Model 60 D, Hewlett-Packard, F. L. Moseley Division, Pasadena, Calif.). The logarithmic output was then fed into an X-Y recorder (Model 135A, F. L. Moseley Division). Log-log (2 X 3 cycles) chart paper (No. 46-7320, Keuffel and Esser) was used. The output of the logarithmic converter was matched to the photometer output by use of the 5 db/in. scale factor of the logarithmic converter and the variable voltage (10 mv/in. maximum) Y axis and Y axis zero adjust of the X-Y recorder. This gave a 2-cycle recording of the photometer oulput-Le., 100 on the photometer = full scale on the recorder, 10 on the photometer = half-scale on the recorder, and 1 ton the photometer = the origin on the recorder. The time-base of the X-Y recorder was set to give the desired trace (the 3 cycles of the X-axis of the chart paper were not used-a linear coordinate X-axis would be ideal, but such 2-cycle log-linear paper is not available). The chopper attenuator was adjusted so that the maximum input to the logarithmic conveiter was about -10 db with respect to 1 volt. The linearity of the logarithmic converter was measured by use of an oscilloscope (Model 545A, Tektronix) with Type L plug-in preamplifier in the d.c. mode and a Polaroid camera attachment (Mode! C 12, Tektronix). The logarithmic converter was found to respond linearly for decay times as short as 0.2 second. The response of the overall system was checked by use of simple R C networks and also found to be linear with decay times as short as 0.2 second. Therefore, in a two-component system the slowest decaying species must have a decay time longer than 0.2 second. The noise level in all studies was found to be more than 65 dp below 1 volt a.c. (6) American Instrument Co., Inc., Instruction Manual No. 838,
Silver Spring, Md.
Reagents. Recrystallized reagent-grade L-tryptophan and L-tyrosine, NBS standard benzoic acid (No. 39 g), and reagent-grade benzaldehyde were used in the studies performed. Stock solutions (10F4M) of tryptophan and tyrosine were made up in 5 X 10-3M sodium methoxide in absolute ethanol purified as previously described (7). Stock solutions ( l O - 3 M ) of benzoic acid and benzaldehyde were prepared in purified absolute ethanol. All solutions of these mixtures were prepared by successive dilution of the stock solutions with the appropriate solvent. The basic solvent was used for tryptophan and tyrosine to improve solubility and to ensure linear decay for the tyrosine (the phenolate ion has a different decay time than the phenolic molecule). Procedure. The phosphorescence signal due to the slow decaying species is measured by extrapolating the linear portion of the logarithmic decay curve (logarithm of phosphorescence signal us. time, t ) at long time, t , after termination of excitation (see Figure 2) to the time corresponding to termination ( t = 0). The phosphorescence signal due to the fast decaying species is measured by subtracting the signal due to the slow decayer from the steady state signal due to the mixture (at t = 0). The wavelengths for excitation of the sample and for measurement of the emission are chosen to yield a logarithmic decay curve giving a linear decay for the slow decaying species over half to two thirds of the Y axis of the recorded plot. The concentration of the components present in the mixture are determined by comparison with analytical curves of each component present in the mixtures. Analytical curves must be determined at each set of wavelengths being used. Measured phosphorescence signals of each component of the two-component mixture must be corrected for attenuation of the excitation intensity. The absorption correction is needed because one species (designated B ) effectively reduces the intensity of exciting radiation for the other species (designated A ) if its (B) absorbance at the wavelength of excitation is sufficient. Therefore, the effective phosphorescence signal of component A , IA*in a mixture containing A and B is given by
(3) and the effective phosphorescence signal of component B, IB* in the same mixture is given by
(4) where the It's are the phosphorescence signals of A or B measured from the logarithmic decay curve of the mixture, and the 5's are the absorption correction factors. The effective phosphorescence signal, I*, is the signal obtainable from the pure component at the same concentration. This correction allows the analyst to use the analytical curves prepared using standard solutions of the pure components. The correction factors, [A and 48, are given ( 5 ) by
and
If the correction factors are less than 1.01, such a correction will not be significant compared to other indeterminate errors and will not be necessary-i.e., EA = € E = 1. The molar absorptivity, E, of each component at the wavelengths in concern were measured by ultraviolet absorption spectrometry. The path length, b, of the Aminco sample cells (7) J. D. Winefordner and M. Tin, Anal. Chim. Acta, 31, 239 (1964). VOL. 39, NO. 4, APRIL 1967
501
4 Figure 2. A . Logarithmic decay curve for benzoic acid (3.00 X 10-4M,T = 2.4 seconds) -benzaldehyde (6.45 X 10-5M,T < 0.1 second) B. Logarithmic decay curve for tryptophan (6.04 X 10-6M,T = 6.4 seconds)-and Tyrosine (5.11 X l O - W , T = 1.4 seconds)
was measured, corrected for reflection losses, and found to be 0.15 cm. The concentrations CA and CB of the mixture were first approximated using no correction ( E A = EB = 1.00). Using the approximated values of CAand CB,values of CA and tBwere calculated. The values oft^ and ( B were then used to obtain corrected phosphorescence signals (Equations 3 and 4) which were used to obtain a second approximation of C A and CB. Further approximations were not warranted due to the inherent errors in phosphorimetryLe., relative errors in the concentrations were less than 5 % after the second approximation.
RESULTS AND DISCUSSION In Figure 2 logarithmic decay curves are given of mixtures of benzoic acid and benzaldehyde and of tryptophan and tyrosine, respectively. The curves show the influence of decay
Mixture no. I (4.65)aqc I1 (4.95)OPC 111 (4.93)590 IV (0.16)o~d
Mixture no.
Added, M 3.00 x 3.08 x 3.08 x 3.29 X
10-4 10-4 10-4 10-6
Added, M
times on the shapes of the recorded logarithmic decay curves (decay times of benzoic acid and benzaldehyde are 2.4 and