linearity is fairly large (-lo3). Compared with data of tryptophan in a solvent without NaI, the analytical linearity range has been increased by a t least an order of magnitude because of the decrease of the detection limit at the low concentrations range. Because it was observed that the compounds could be gradually decomposed by the strong basic solvent, all the measurements were performed using freshly-prepared samples. Limits of Phosphorimetric Detection. Table I reports the limits of phosphorimetric detection of a wide variety of compounds of biological importance. The sodium iodide heavy-atom enhancement factor IpN"'/Ip is defined as the ratio of the molar phosphorescence signal in 1 M NaOH-1 M NaI (1:l) solution to that in 1M NaOH aqueous solution, respectively, and was deduced from the following equation:
where ip*, ,'?i c p ' are, respectively, the phosphorescence signals (in A) of the sample, S, the reference blank, R, and the concentration (mol l.-I) of the substance, S, under investigation in a solution of NaOH-NaI; and the other parameters correspond to those items for only NaOH solution (no NaI). Except for 6-methylmercaptopurine, the heavy-atom enhancement factor was found for most compounds to be greater than unity. It ranged from 1.0 for 6-chloropurine to 12 for guanine, and 40 for tryptophan. The data show that the heavy-atom effect on room temperature phosphorescence varies appreciably with the molecular structure of the compounds and the nature of the excited states involved. Advantages of R T P f o r Analysis. An obvious advantage of R T P technique using paper for analytical purposes, besides its simplicity, is the small amount of sample which is required for every measurement; i.e., only 5 ~1 is usually needed to prepare a sample, whereas low temperature phosphorimetry usually requires 100 ~1 to 1ml and conventional fluorimetry requires as much as 4 ml. Therefore, the absolute sensitivity (in ng) of R T P is in some cases much higher than the con-
ventional techniques; for tryptophan, an absolute limit of detection of 0.15 ng can be achieved. The significant enhancement of tryptophan phosphorescence is of special interest in protein analysis since most proteins have been found to emit via their tryptophan residues. Para-aminobenzoic (PABA) acid which has a very low limit of detection (9 X mol l.-l) is also of medical interest because it has been used to raise blood salicylate level, or in the laboratory as a sulfonamide antagonist, or more commonly as a sunscreen agent (14). The absolute limit of phosphorimetric detection of PABA is 0.06 ng. This study has shown that the external heavy-atom effect on room temperature phosphorescence is applicable to a wide variety of compounds of biological importance. In some cases, the sensitivity of phosphorimetric detection is increased by several orders of magnitude. It is therefore suggested that the use of the external heavy-atom effect be applied to improve the sensitivity of room temperature phosphorimetric method of analysis. This technique, which is a rapid, simple method of analysis, would be quite suitable for clinical analysis. LITERATURE CITED (1) C. M. O'Donnell and J. D. Winefordner, Ciin. Chem., ( Winston-Salem,N.C.), 21, 285 (1975). (2) J. J. Aaron and J. D. Winefordner, Talanfa, 22, 707 (1975). (3) E. M. Schulman and C. Walling, Science, 178, 53 (1972). (4) E. M. Schulman and C. Walling, J. fhys. Chem., 77, 902 (1973). (5) P. G. Seybold, R. K.Sorrel1 and Schuffert, 165th National Meeting, American Chimical Society, Dallas, Texas, April 13, 1973. (6) S. L. Wellons, R. A. Paynter, and J. D. Winefordner, Spectrochim. Acta, fartA, 30, 2133 (1974). (7) P. G. Seybold and W. White, Anal. Chem., 47, 1199 (1975). (8) M. Kasha, J. Chem. fhys., 20, 71 (1952). (9) T. Medinger and F. Wllkinson, Trans. Farad. Soc., 61, 620 (1965). (10) R. F. Chen, Anal. Biochem. Biophys., 144, 552 (1971). (11) G. D. Boutilier, C. M. O'Donnell, and R. 0. Rahn, Anal. Chem., 46, 1508 (1974). (12) R . F. Chen, Amhco Fiuorescence News/., 9, No. 2 and 3, 9 (1975). (13) B. A. Baldwin and H. W. Otfen, J. Chem. fhys., 49, 2933 (1968). (14) The Merck Index, Merck and Co., Inc. (1968 ed.) p 53.
RECEIVEDfor review February 17, 1976. Accepted April 7, 1976. This work solely supported by the U.S. Public Health
Comparisons between the Luminol Light Standards and a New Method for Absolute Calibrations of Light Detectors Paul
R. Michael and Larry R. FaulkneP
Department of Chemistry, University of Illinois, Urbana, Ill. 6 180 1
An actinometric technique for calibrating detectors used in the measurement of absolute light levels has been developed. It features a flexible light guide, which transmits monochromatic light from a mercury-xenon lamp to a sample cell contained in an integrating sphere. Measuring the quantum intensity from the guide, both actlnometrlcally and with a photomultiplier situated at the sphere's viewing port, allows one to calculate the absolute photometric calibration factor. The procedure also includes a relatively simple means for checking the wavelength dependence of the callbration factor. The results of the calibration were found to be consistent with the published liquid light standards based on the luminol chemiluminescent reactions in water and in DMSO. 1188
ANALYTICAL CHEMISTRY, VOL. 48, NO. 8, JULY 1976
The study of luminescent materials frequently is concerned with the quantum efficiency of a light emitting process. Determining it usually requires measurements of the total photon yield of a reaction over the entire wavelength range of its emission spectrum, which can span 100 nm or more (1,2).A sensitive detector such as a photomultiplier tube is usually required to detect the low light intensities generated in most luminescence experiments, and it must be calibrated in absolute units (e.g., incident photons per microcoulomb of integrated output current), so that the number of photons generated can be extracted from easily measured parameters. The calibration is complicated by the requirement that it be known for an extensive wavelength range in order to accom-
modate the entire luminescence spectrum. Such absolute light measurements are, therefore, different from and inherently more difficult than the relative measurements typically encountered in analytical chemistry. We will describe below a relatively simple and straightforward means, based on the ferrioxalate actinometer (3), of calibrating a light detector in absolute terms. An added feature of the technique is the ease with which the wavelength dependence of the calibration factor can be obtained. Although we have applied it only to a specialized detection apparatus for studies of electrogenerated chemiluminescence, the calibration procedure is flexible enough to be used for other detectors as well. Perhaps of more general interest, we have also compared our calibration data directly to the luminol liquid light standards of Lee and Seliger ( 4 , 5 ) .T h e results of our comparison, together with others appearing in the literature (6, 7), indicate that the luminol reactions in water and DMSO are accurate means for standardizing light measurements.
EXPERIMENTAL Apparatus. The photometric apparatus used in our work is almost identical to the one described by Bezman and Faulkner (8). Only a brief description of it will be given here. The sample is enclosed by an integrating sphere made from a modified 3-1. flask, which is coated on the inside with barium sulfate reflectance paint from Eastman Kodak. The sample cell enters the sphere and is held in position by a standard taper joint attached to the sphere surface. The sample cell itself is identical to the electrochemical cell designed by Bezman (9). A photomultiplier tube faces the sphere’s viewing port and observes the sample luminescence through a rhodamine B quantum counter (8.0 g/l. in ethylene glycol).Essentially the only modifications to the original apparatus are a) the addition of a light baffle necessary to prevent light from leaking around the edges of the quantum counter cell and b) a reduction in photomultiplier gain required by our discovery of an intermittent short inside the photomultiplier between two adjacent dynodes in the middle of the chain. This problem was eliminated by permanently shorting the two dynodes externally. In addition to these changes, the interior of the integrating sphere was stripped and recoated with reflectance paint to avoid any problems arising from aging of the paint surface. The fluorescence instrument used to obtain corrected spectra and used as a light source in the calibration procedure was an AmincoBowman spectrophotofluorometer (SPF) equipped with either a 150-W xenon or a 200-W mercury-xenon lamp and an ellipsoidal off-axis focusing attachment. Its detector was a Hamamatsu R446-S photomultiplier tube with S-20 response. Reagents. Potassium ferrioxalate was prepared from reagent grade potassium oxelate and ferric chloride according to the procedure of Hatchard and Parker ( 3 ) .Luminol was obtained from Eastman and was used as received. The hemoglobin employed in the aqueous luminol experiments was twice recrystallized bovine hemoglobin from Sigma Chemical Company. All other chemicals were of reagent quality. Actinometric Calibration. A block diagram illustrating the equipment used in the calibration of the integrating sphere detector is shown in Figure 1.The 436-nm line of the mercury-xenon arc was isolated by the excitation monochromator and focused on one end of a y8 X 48 in. flexible light guide, which was manufactured for enhanced uv transmittance by American Optical Corporation. The slits were set to 2 mm, which yielded a bandpass of 12 nm. The guide, mounted in a machined brass block placed in the standard cell holder, was used to transmit light emerging from the monochromator t o the integrating sphere. The photon flux from the end of the guide was measured using a 0.15 M ferrioxalate actinometer, in which ferrioxalate ion is photoreduced with a known quantum efficiencyto ferrous ion. The free end of the light guide was immersed in sufficient solution volume (typically 10 ml) that the light emitted there had to travel through slightly more than a 1-cm thickness in every direction. This arrangement ensured that essentially all of the light was absorbed by the actinometer ( 3 ) .After an exposure for an accurately known period, the ferrous ion concentration was measured spectrophotometrically using the 1,lO-phenanthrolinemethod. Producing concentrations of ferrous ion equivalent to an absorbance of 0.1-0.3 for a 1-cm cell at 510 nm required an exposure time of approximately 15 min. The photon output was then calculated by dividing the number of ferrous ions
--.
U n.
or X.
W P M Tub.
Lamp
Flgure 1. Block diagram of equipment used in actinometric calibra-
tions produced by the product of the exposure time and the published actinometer quantum efficiency at 436 nm (3).The only remaining step was to insert the light guide into the sample cell in the sphere and measure the correspondingphotocurrent. Dividing the photon output by the photocurrent yielded a calibration factor of 3.6 k 0.2 X 1013 photons/pC. It should be noted that this factor does not bear any obvious relationship to the one previously determined by Bezman and Faulkner (8)because of the subsequent changes made in the apparatus. This calibration procedure involves the assumption of a constant or nearly constant photon flux from the light guide. In actual practice, arc wander can cause changes in the quantum flux during the 15-min exposure time, especially if an old arc lamp is used as the light source. This effect could be monitored by observing the scattered light from the collecting end of the light guide via the SPF’s own detector as shown in Figure 1.If severe fluctuations occurred,the run was ignored; but for minor ( 600 nm, the sensitivity of the apparatus rises sharply until a peak is reached a t 635 nm, where it is about 70 times as sensitive as in the region below 600 nm. The gradual decrease in sensitivity towards 700 nm results from the decreasing quantum effi-
ciency of the PM tube’s photocathode. Any light passing directly through the quantum counter solution obviously registers much more strongly than light converted to rhodamine B fluorescence. T o calculate the photometric factor a t any wavelength past 600 nm, it is necessary only to divide the corresponding relative sensitivity, S, from Figure 3 into the calibration factor of 3.6 X 1013photons/pC. When studying luminescent systems with spectra extending past 600 nm, one can obtain a specific calibration factor for that particular system from the relation. [JISdXI-l Weighted response (photons/&) = (3.6 X where I describes the corrected emission spectrum normalized to unit area, i.e., JZdX = 1. ‘Comparison with t h e Luminol Light Standards. Considerable interest has arisen in recent years over the development of liquid light standards. The availability of a solution phase light source having a known quantum intensity or a total photon yield would greatly facilitate both the absolute calibration of light detectors and the comparison of luminescent yield data obtained by different laboratories. The well known chemiluminescence of luminol has been suggested as such a light standard (4).It can be oxidized in either water or DMSO to yield a photon emission which is 99% complete in 4 to 5 minutes as all of the luminol in solution is consumed. Under specified conditions, Lee and Seliger ( 4 , 2 2 )reported that 1ml of luminol solution having an absorbance of 1.0 cm-’ a t 347 nm in water or 359.5 nm in DMSO emits a total of 9.75 f 0.7 X 1014 photons in either solvent. In order to relate our calibration data t o these standards, the luminol reactions were run in the integrating sphere detector. Applying the calibration factor of 3.6 X 1013photonsf FC directly to the integrated photocurrent provided apparent luminol photon yields of 11.7 f 0.6 X 1014 €or the aqueous reaction and 34 f 3 X 1014for the DMSO reaction. These are given in Table I as uncorrected results. It appears that the discrepancies between our initial findings and those of Lee e t al. result from small tails of the emission spectra which extend past 600 nm. The effect is more severe for the reaction in DMSO, because its emission is a t larger wavelengths than that of the reaction in water. The emission peaks are a t 424 nm for the aqueous system and a t 485 nm for the DMSO case. Corrected emission spectra (12, 13) show that more than 95%of the luminol emission in either solvent lies within the region of flat response for our detector. However, photons passing through the quantum counter contribute disproportionately to the photocurrent, as evidenced by the response curve in Figure 3; thus the longwavelength tail will cause luminol to appear to emit a greater number of photons than it actually does. A series of experiments was performed with filters to determine the fraction of the photocurrent produced by photons not absorbed by the quantum counter. A red sharp-cut filter, Corning C. S. 2-62, was inserted between the integrating sphere exit port and the rhodamine B cell. The filter was selected so that its short-wavelength cutoff would be slightly to the blue side of the rhodamine B cutoff. For comparison, the rhodamine B solution (8 g/lJ has an absorbance of 1 cm-l a t 623 nm, while the filter has a unit absorbance a t approximately 600 nm. Thus, the filter transmits a slightly greater portion of the luminol spectrum than the quantum counter but, with the filter in place, the only significant contribution to the photocurrent comes from the long wavelength emission passing directly through the rhodamine B, because of the greatly enhanced sensitivity in that region. T o compare this current to the photocurrent produced by the total luminol emission, the C. S. 2-62 filter was replaced with a Corning C. S.0-52 sharp-cut filter, which was selected to transmit the entire luminol spectrum with a short wavelength cutoff a t 350
Table I. Luminol Photon Yielda Solvent Uncorrected Hz0 11.7 f 0.7 X 1014 DMSO 34 f 3 x 1014
Fraction passing C. S. 2-62 Filter 24.5%
Corrected 8.8 f 0.5 X 1014 68.7% 10.6 I 0 . 9 X 1014 a For 1 ml of a luminol solution having an absorbance of 1.0 cm-l at 347 nm in HzO or 359.5 nm in DMSO, nm. The purpose of this second filter was to maintain the losses of light reflected off the faces of the first filter and the slightly different geometry of the integrating sphere apparatus caused by insertion of the filter. T o ensure that the C. S. 0-52 filter transmitted all of the luminol emission, it was replaced by a glass microscope slide (cutoff