Analyst, December, 1983, Vol. 108, Pe. Id7l-ll76 147 1 Low-tem perature Luminescence Spectroscopy Using Conduction Cooling and a Pulsed Source Luminescence Spectrometer Alun T. Rhys Williams and Stephen A. Winfield and James N. Miller Pevkin-Elmer Limited, Beaconsfield, Buckinghamshire, H P 9 1QA Department of Chemistvy, Lozcghborough University of Technology, Loughborough, Leicestershire, LE11 3TU A new conduction cooling device for low-temperature luminescence measure- ments is presented and its use evaluated in a pulsed source luminescence spectrometer, Fluorescence and phosphoresence quantum yields are reported for some polyaromatic hydrocarbons a t 77 K. Keywords : Low-temperature luminescence spectroscopy ; conduction cooling ; pulsed souwe Low-temperature fluorescence and phosphorescence spectroscopy are not popular assay methods amongst analytical chemists.Potentially these techniques offer both increased sensitivity and selectivity of analysis. The latter is obtained by spectral line narrowing at low temperatures and by the use of time-resolved phosphorimetry. Increased sensitivity is predominantly observed for those compounds which phosphoresce rather than those which have relatively high fluorescence quantum yields, Phosphorescence detection limits range from 0.0004 to 4 pg ml-l for polycyclic aromatic hydrocarbons, whilst drugs such as pheno- barbital and procaine have been detected at 7 and 2.5 ng ml-l, respectively. The reluctance to use low-temperature luminescence spectroscopy probably arises from the problems associ- ated with the need to cool samples to liquid-nitrogen temperature, 77 K.Conventional low- temperature accessories use immersion cooling, i.e., the sample in a synthetic fused silica tube is cooled by immersion in liquid nitrogen held in a Dewar flask. Two major problems arise from this method ; poor precision resulting from irreproducible sample positioning and erratic freezing rates and very slow sampling rates, Additional problems include a gradual decrease in signal intensity due to an accretion of ice crystals in the base of the Dewar and a build-up of ice on the outside of the base of the Dewar if the sample compartment has not been adequately purged with dry nitrogen. One other factor that can deter potential users is the relatively high cost of phosphorescence accessories.In addition to a costly synthetic fused silica Dewar some means of mechanically chopping the excitation and emission beams, i.e., a rotating can with its associated motor and control module, is required. Attempts have been made to improve the precision of low-temperature measurements by rotating the sample tubel ; this averages any inhomogeneities in the frozen sample but does not overcome positioning errors. Conduction cooling provides an alternative to immersion cooling as a means of improving the precision and speed of analysis. Ward and co-worker~~~~ discussed two conduction cooling systems. In the first, a copper rod was immersed in liquid nitrogen and the sample was cooled by contact with the rod. Although the speed of analysis was improved by a factor of three over conventional phosphorimetry, the detection limits were similar. Measurements indicated that the operating temperature of the system was 100 K.In the second system the copper rod was cooled by flowing through it liquid nitrogen, the operating temperature being 85 K. In this instance detection limits were lower, though the main advantage was the improved pre- cision and increased speed of analysis. In this paper we describe a conduction cooling device using a high-purity copper rod immersed in liquid nitrogen used in conjunction with a pulsed source luminescence spectro- meter.4 The advantages of using a pulsed source are as follows: increased sensitivity, as a pulsed source produces higher peak intensities in the ultraviolet region than a continuously operated xenon-arc lamp with a mechanically chopped excitation beam; better time resolution , which permits analysis of phosphors with short lifetimes (0.1-50 ms)596 in the presence of1472 RHYS WILLIAMS et aZ.: LOW-TEMPERATURE LUMINESCENCE Analyst, VOZ. 108 relatively long-lived phosphors ; and the ability to discriminate fluorescence and scattered light from long-lived phosphorescence without the use of a mechanical phosphoroscope. The events that occur during excitation of a sample with a pulsed source are shown in Fig. l.'p8 As the observed phosphorescence signal (P) represents only a fraction of the total phosphorescence intensity (PT), the area under the corrected phosphorescence-emission spectrum has to be multiplied by a factor that is related to the characteristics of the pulsed- source electronics and the lifetime of the phosphorescent speciesg The excitation flash width (if) is assumed to be much smaller than the phosphorescence lifetime, the latter being ca.0.5 x 10-3-10 s for organic phosphors. -t*-fg-* Time+ Fig. 1. Events occurring during the excitation of a sample with a pulsed source. A, 'lhc excitation pulse; 73, build-up of luminescence signal I, and then exponential decay; tp, width a t half peak height; t d . delay from beginning of pulse to beginning of observation; t,, gate width of detector. The total observed decay for a single exponential may be expressed by the following equa- tion : where PT is the total phosphorescence, yo is the intensity at zero time and r is the single- exponential decay time.This may be evaluated as The integrated phosphorescence intensity, P, during the time interval tg, at a time td after to is given by the following expression : The fraction of light observed is then given by the ratio of equations (2) and (3) : Equation (4) holds for a single pulse of excitation light; for a pair of pulses with a cycle time of 20 ms, equation (4) becomesDecember, 1983 USING CONDUCTION COOLING AND A PULSED SOURCE 1473 where all times are in microseconds. Experimental Low-temperature fluorescence and phosphorescence measurements were obtained on a Perkin-Elmer, Model LS-5, luminescence spectrometer fitted with a Hamamatsu R928 red- sensitive photomultiplier. Data were recorded on a Perkin-Elmer, Model 3600, Data Station.The spectrometer uses a 9-W xenon lamp pulsed at line frequency, tlie pulse width at half peak height being less than 10 ps. The signals from tlie sample and reference photomultipliers are gated by the microprocessor-controlled electronics and measured by an A-D converter with an 18-bit dynamic range. The reference photomultiplier forms part of the quantum-corrected reference system, which enables corrected excitation spectra to be obtained. In the fluor- escence mode the signals from the sample photomultiplier are gated for the duration of the flash whilst in the phosphorescence mode the gating is delayed so that it no longer coincides with the flash. The discrimination between fluorescence emission and long-lived phosphorescence is difficult to achieve on instruments using a d.c.source, e.g., a 150-W xenon lamp. Fig, 2(a) shows the low-temperature “fluorescence” spectrum of coronene in hexane obtained using a d.c. operated 150-W xenon source. Fig. 2(b) sliows the fluorescence spectrum obtained using a pulsed 9-W xenon source. The difference results from the fact that the peaks above 500 nm are phosphorescence transitions. When using a pulsed source the sample photomultiplier can be gated to look at tlie signal at the instant of the flash and again just before the next flash. By subtracting the signals a fluorescence signal free from any long-lived phosphorescence and dark current is obtained. 80 60 8 e g 4c 2 m 2c 100 80 60 40 20 400 500 600 0 500 Wavelengt hlnm Fig. 2. (a) Low-temperature fluorescence-emission spectrum of coronene in hexane measured with a d.c. operated 150-W xenon lamp.(b) Low-temperature fluorescence-emission spec- trum of the same sample measured with a pulsed 9-W xenon lamp. A !I I I# I I I I I I c l I 2 ‘D Fig. 3. Diagram of the copper conduction cooling rod. A, Sample; B, optical window ; C, hollow cavity; D, hole allowing nitrogen to boil over the sample tubes.1474 RHYS WILLIAMS et d. LOW-TEMPERATURE LUMINESCENCE Aftdyst, VOZ. 108 The sample compartment of the LS-5 was changed to incorporate the conduction cooling low-temperature accessory. The conduction rod was machined from high-purity 99.99% copper with a thermal conductivity of 4.05 J s-l cm-l K-l (Fig. 3). Sample tubes of 4 mm 0.d. and 2 mm i.d. were inserted into a hole drilled in the top of the copper rod with a slot milled to 20 mm deep from the end to serve as the optical windows.The rod is hollowed leav- ing a cavity of 12.7 mm i.d. extending from the open end. A hole is drilled into the cavity allowing nitrogen gas to flow around the sample tube in the slot. Once sealed in position in the sample compartment the flow of nitrogen serves to keep the slot free of moisture from the surrounding air. The bottom part of the copper rod is immersed in 2 1 of liquid nitrogen held in a box constructed of stainless steel, insulated by expanded polystyrene. The accessory is enclosed by the sample compartment cover through which the sample tubes are inserted by means of a removable lid. The inside walls of the windows of the optical module were kept free of ice by a stream of dry nitrogen at a flow-rate of 1 1 min-l. Topping up with liquid nitrogen is necessary every 15-20 min, depending upon the external air temperature.The calculation of the quantum yields was greatly simplified by the use of a desk-top computer interfaced with the spectrometer, with the observed and factored phos- phorescence-emission areas being automatically calculated. Reagents Methylcyclohexane (spectroscopic grade) was fractionated and stored over concentrated sulphuric acid for 1 week. The solvent was then washed with de-ionised water and again fractionally distilled. This process, which was also applied to cyclohexane, greatly reduced the background fluorescence. Methylcyclohexane was preferred as the solvent as it formed a clear glass at 77 K, whilst cyclohexane formed a snow that increased light scattering and hence the background signal.Aqueous solutions require a thicker walled sample tube to withstand the expansion when cooled to 77 K. Stock solutions of the chemicals were prepared and stored at 4 "C. Spectra were recorded with a 2.5-mm spectral band pass and were corrected for instrument response from 250 to 630 mm. Results and Discussion Confirmation of equation (5) for calculating the phosphorimeter factor was obtained in the following way. A transferrin - terbium complex (loA6 and 4 x 10W M, respectively) at pH 8.0 was excited at 280nm and the area under the corrected emission spectrum of terbium measured at various delay ( t d ) and gate (tg) times (Table I). The observed areas were multi- plied by the appropriate factor to produce a mean result of 2513 with a relative standard deviation (RSD) of & 2.03%.The graph shows that almost all the signal from short-lived phosphors is measured whilst only a fraction of a long- lifetime emission is measured. For the same quantum efficiency, short-lived phosphors A T value of 1.22 ms was used. The variation of P,,P for various values of T is shown in Fig. 4. TABLE I CALCULATION OF TOTAL PHOSPHORESCENCE EMISSION FROM OBSERVED AREAS AT VARYING id AND t g USING EQUATION (4) T = 1.22 ms. td/ms 0.03 0.1 0.5 1.0 2.0 3.0 0.1 0.1 0.1 tglms 0.5 0.5 0.5 0.5 0.5 0.5 1 .o 2.0 3.0 0 bserved emission area 823 79 1 596 359 169 78 1286 1851 2087 Total phosphorescence emission* 2508 2553 2570 2423 2589 2512 2495 2493 2477 *Mean = 2513; S.D.50.95 (n = 9).December, 1983 USING CONDUCTION COOLING AND A PULSED SOURCE 1475 (7 ca. s) will show an order of magnitude increase in intensity compared with a long-lived species (T ca. 10-1 s). Fluorescence quantum yields were calculated as described by Rhys Williams et aZ.1° assuming a fluorescence quantum yield of 1.00 for 9,lO-diphenylanthracene at 16 12 % cl‘ 8 - - - ~ 8 8 8 8 8 1 10 100 1000 10000 Time/ms Fig. 4. Variation of PT/P (ratio of the total phosphorescence, PT, to the integrated phosphores- cence intensity, P, for different T values (ms) measured with a t d of 0.1 ms and t , of 1.0 ms. 77 K.ll Because a 2 mm i.d. cuvette is used in the phosphorescence accessory, solute con- centrations of up to five-fold higher were used compared with fluorescence measurements made using 10 mm path length cuvettes. This enabled good quality spectra with high signal to noise ratios to be used in the calculation of quantum yields.The observed peak intensity of the 9,lO-diphenylanthracene increased approximately five- fold on going from 293 to 77 K. This increase was predominantly due to band sharpening, 250 350 450 5150 Wavelengthlnm Fig. 5. (a) Room-temperature excita- tion and emission spectra of 9,lO-diphenyl- anthracene in methylcyclohexane. (b) Low-temperature (77 K) excitation and emission spectrum of the same sample.1476 RHYS WILLIAMS, WINFIELD AND MILLER although an increase in the absorption coefficient cannot be dismissed. Fig. 5 compares the room temperature and 77 K excitation and emission spectra of 9,lO-diphenylanthracene.The results of the quantum yield determination are shown in Table 11. The reproducibility of the quantum yield values at 77 K is & S-lO% (RSD). Fluorescence quantum yields agreed, within experimental error, with those in the literature with the exception of fluorene which Parker and Hatchard12 assumed to have a quantum yield of 0.54 at room temperature and at 77 K. Phosphorescence quantum yields, with the exception of benzene, were also in good agreement with previously determined values. The discrepancy with the benzene is probably due to the level of impurities, as both the quantum yield and lifetime differ signifi- cantly. TABLE I1 LOW-TEMPERATURE FLUORESCENCE AND PHOSPHORESCENCE QUANTUM YIELDS (Qf AND QP) RELATIVE TO ~,~O-DIPHENYLANTHRACENE AT Qf = 1.00 Values in parentheses are literature values.Aex. rangel Compound Solvent Aex./nm nm Lifetimels Observed Qr Observed Qp References Anthracene .. .. MCH* 252 360-490 N.0.t 0.29 (0.27) N.o. 12 Benzene .. . . .. MCH 254 265-340 5.8 ( 8 ) 0.20 (0.21) 0.31 (0.19) 1 2 BenzoDhenone .. .. MCH 250 400-500 6 x N.o. 0.89 (0.84) 13 . . (8 x 10-5) Coronene .. .. .. MCH 303 350-510 9 0.12 0.07 Fluorene . . .. . . MCH 265 275-380 N.o. 0.77 (0.54) N.o. (0.07) 12 Naphthalene .. Hexane 276 300-400 N.o. 0.35 (0.39) N.o. (0.008) 12 Tetraphenylbutadfene . . MCH 346 350-520 N.o. 0.90 N.o. Triphenylene . . . . Hexane 259 330-440 10.5 (15) 0.04 (0.06) 0.25 (0.28) 1 2 Eu(TTA),$ .. .. Ethanol 275 3.6 x 10-4 N.O. 0.13 (0.18) 14 * MCH, methylcyclohexane.t N.o., not observed. $ TTA = thenoyltrifluoroacetonate. Results obtained a t room temperature, 25 “C. No special technique or practice was required in order to maintain reasonable precision using After initial cool-down it was only necessary to add liquid Samples were cooled and reached maximum phos- the conduction cooling device. nitrogen on average at the rate of 1 1 h-l. phorescence intensity 3045 s after insertion. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Lukasiewicz, R. J., Mousa, J. J., and Winefordner, J . D., Anal. Chem., 1972, 44, 1339. Ward, J . L., Bateh, €3. P., and Winefordner, J . D., Appl. Spectvosc., 1980, 34, 15. Ward, J . L., Walden, G. L., Bateh, R. P., and Winefordner, J. D., Appl. Spectrosc., 1980, 34, 348. Rhys Williams, A. T., I n t . Lab., 1981, 11, 90. Harbaugh, I<. F., O’Donnell, C. M., and Winefordner, J. D., Anal. Chem., 1971, 45, 381. Harbaugh, K. F., O’Donnell, C. M., and Winefordner, J. D., Anal. Chem., 1974, 46, 1206. Fisher, R. P., and Winefordner, J. D., Anal. Chem., 1972, 44, 948. O’Haver, T. C., and Winefordner, J. D., Anal. Chem., 1966, 38, 1258. Boutilier, G. D., Bradshaw, J. D., Weeks, S. J., and Winefordner, J. D., Appl. Spectrosc., 1977, 31, Rhys Williams, A. T., Winfield, S. A., and Miller, J. N., Analyst, 1983, 108, 1067. Mantulin, W. W., and Huber, J. R., Photochem. Photobiol., 1973, 17, 139. Parker, C. A., and Hatchard, C. G., Analyst, 1962, 87, 664. Gilmore, E. H., Gibson, G. E., and McClure, D. S., J . Chem. Phys., 1955, 23, 399. Baumik, M. L., and Telk, C. L., J . o p t . SOG. A m . , 1964, 54, 1211. 307. Received July 4th, 1983 Accepted July 27th, 1983