Molecular fluorescence measurements with a charge-coupled device

a charge vs time curve for reduction and reoxidation of. Rh2(ap)4Cl underthe same conditions and the curve shows only a flat line when the potential i...
0 downloads 0 Views 1MB Size
282

Anal. Chem. 1989, 6 1 , 282-285

c). The plot of current vs time-'I2 gives a straight line that levels off a t a current intensity of 145 pA (Figure 4b). The steady-state current is largely due to the catalytic reduction of O2 by the dirhodium complex since the direct reduction of O2 on a blank electrode under the same conditions shows a residual current of only 20 FA. As the electrolysis potential is stepped to +0.3 V, no anodic current is obtained (curve c', Figure 3). This behavior indicates the irreversible nature of the catalytic reduction of 02.The insert of Figure 3 illustrates a charge vs time curve for reduction and reoxidation of Rh2(ap),C1 under the same conditions and the curve shows only a flat line when the potential is stepped from -0.4 to +0.3 V (the charge due to the oxidation of the reduced Rh2(ap),C1 is negligible). The increased electrolysis current using the large electrode also allows the easy and accurate analysis of Hz02produced by the electrode reaction. The enrichment of H202in the buffer solution determined by an iodometric method ( I 7) reveals a solution concentration of 3.5 x mol/L after electrolysis for 4 h which gives a 20% current efficiency. The low current efficiency is probably due to decomposition of much of the generated Hz02over the 4-h period. The current-time curves of the electrode containing the Rh2(ap),C1 complex show a steady increase after each double-potential-step scan under Ar or OF This increase is clearly due to the surface enrichment of the dirhodium complex after each cycle which results from the insolubility of both the oxidized and reduced form of the dirhodium compound in water. In summary, this study shows that the mechanical strength of a large surface area carbon paste electrode can be improved through the described honeycomb design. The electrode can

be easily used for bulk solution electrolysis and allows the convenient analysis of the soluble components generated from the electrode reaction. This type of electrode can also be used to evaluate the chronoamperometric, chronocoulometric, and chronopotentiometric behavior of electroactive substances doped in carbon paste electrodes and be a valuable tool for studying the kinetics and efficiency of electrocatalysis.

LITERATURE CITED Bear, J. L.; Yao, C.4.: Cacdevielle, F. C.: Kadish. K. M. Inom. Chem. 1988, 2 7 , 3782. Korfhage, K. M.; Ravichandran, K.; Baldwin, R. P. Anal. Chem. 1984, 5 6 , 15-14. Takeuchi, E. S.; Murray, R. W. J. Electroanal. Chem. 1985. 768, 49. Castro, M. D. L.; Valcarcel, M.; Albahadily, F. N.; Mottola, H. A. J . Electroanal. Chem. 1987, 219, 139. Halbert, M. K.; Baldwin. R. P. Anal. Chem. 1985, 5 7 , 591. Geno, P. W.; Ravichandran, K.; Baldwin, R. P. J. Electroanal. Chem. 1985, 783, 155. Kamin, R. A.; Wilson. G. S. Anal. Chem. 1980, 52, 1198. Ravichandran, K.: Baldwin, R. P. J . Elecfroenal. Chem. 1981, 726, 293. Bear, J. L.; Liu, L.-M.: Kedish, K. M. I n w . Chem. 1987, 26, 2927. Bard, A. J.; Faulker, L. R. ElectrochemicalMethods; Wiley: New York, 1980; p 143. Tyrrell, H.J. V. Dlffusbn and Heat Flow In Liquids; Butterworths: London, 1961; p 138. Lacaze, P. C.: Minh, C. P.; Delarnar, M.; Dubois, J. E. J. Elecfroanal. Cham. W80. ..., 708. .- 9. Chin, J. W.; Georges, J. J. Electrottnal. Chem. 1986, 270, 205. Georges, J.; Desrneftre, S. Electrochlm. Acta 1984, 2 9 , 521. Ravichandran, K.; Baldwin. R. P. Anal. Chem. 1983, 5 5 , 1586. Murray, R. W. I n phvsicel Methods of Chemlstry; Rossiter, B. W., Hamllton, J. F., Eds.; Wiley: New York, 1986; Vol 12, p 557. Vogel, A. I. A Text-Book of Quanflfaflve Inorganic Analysls; Wiley: New York 1961; Vol 59, p 363.

.

RECEIVED for review July 6,1988. Accepted October 31,1988. The authors thank the Robert A. Welch Foundation (Grant No. E-918) for financial support of this work.

Molecular Fluorescence Measurements with a Charge-Coupled Device Detector Patrick M. Epperson, Rafi D. Jalkian, and M. Bonner Denton* Department of Chemistry, University of Arizona, Tucson, Arizona 85721 The sensitivity of a fluorescence measurement depends upon a number of factors including source intensity and stability, excitation and emission collection efficiency, background fluorescence from solvent and cuvettes, and the detector sensitivity. In the last decade, the majority of work aimed at increasing sensitivity focused on the development of more intense excitation sources. The most significant results were achieved by the use of laser excitation sources (1-3). Whereas impressive limits of detections were achieved by such systems, the cost and complexity of such sources make them impractical for use in most routine analytical instrumentation. Additionally, the use of powerful excitation sources can result in sample heating and photodecomposition. An alternate method of increasing the sensitivity of a fluorometer involves the more efficient detection of the available fluorescence emission from the sample by increasing the detector efficiency. Multichannel array detectors are actively being investigated as a means of improving upon the sensitivity of the single-channel photomultiplier tube (PMT). One- and two-dimensional array detectors such as vidicons, photodiode arrays, and intensified photodiode arrays are being investigated for use in many areas of atomic and molecular spectroscopy (4-8). These detectors have a demonstrated multichannel advantage over PMT-based instrumentation resulting in improved signal-to-noise (S/N) ratios in situations 0003-2700/89/036 1-0282$01.50/0

requiring the measurement of a complete spectrum in a short period of time (9-12). However, these detectors suffer from relative high detector noise and dark current levels that limit their sensitivities when measuring low light level signals. Currently, the P M T is still the detector of choice when measuring extremely low photon fluxes of less than a few thousant photons at a single fixed wavelength for quantitative fluorescence spectroscopy. Another type of multichannel array detector, the chargecoupled device (CCD), shares the same multichannel advantages as the above detectors but has a much lower detector noise level and dark current (13-15). The low detector noise and the ability to integrate photogenerated charge for several minutes make the CCD as sensitive as a photon-counting PMT for measuring extremely low light levels. Additionally, the CCDs ability to combine photogenerated charge packets from adjacent detector elements via a process known as binning allows the performance of the detector to be tailored to the needs of the spectroscopic measurement (14,16). One- and two-dimensional CCDs were evaluated for use in UV-vis absorption, atomic emission, and Raman spectroscopy; however, to date CCDs have not been evaluated as detectors for quantitative molecular fluorescence. Given both the multichannel advantages of the CCD together with the ability to detect low photon fluxes that is equal or superior to PMTs, 0 1989 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 61, NO.

rl computer

Figure 1. Spectrograph constructed for the molecular fluorescence measurements employing a 512 by 320 element CCD detector. we felt that such an evaluation would be useful. This paper describes the design and evaluation of a fluorometer employing a charge-coupled device (CCDJarray detectur for quantitative molecular fluorescence spectroscopy. The primary goal of this study is to develop a practical instrument for the measurement of steadystate fluorescence measurements without resorting M the use of intense excitation sources. This approach is realized by the use of a mercury pen lamp as a fluorescence excitation source that is modest in both cost and photon flux. The sensitivity and dynamic range of the instrument were evaluated by measuring solutions of anthracene in ethanol M. solvent at concentrations ranging between lo4 and Anthracene was chosen because it is commonly used to measure the sensitivity of fluorescence instruments allowing the results of this work to be compared against results from previous studies employing different detectors and excitation sources. The distinctive fluorescence spectrum of anthracene also aids in evaluating the spectral resolution of the fluorometer. EXPERIMENTAL SECTION The instrument constructed for the molecular fluorescence measurements is mechanically simple and reliable. The design

3. FEBRUARY 1, 1989

283

emphasizes the efficient use of all the spectral information availahle from the excitation of the analyte molecules in ethanol by simultaneously measuring the fluorescence emission as well as the elastically and Raman scattered light from the solvent. The scattered light intensities are used to correct for source and detector drift between the sample and blank measurements. Fluorescence Spectrograph. A diagram of the fluorescence spectrograph built for these studies is shown in Figure 1. The 254.line of the mercury.penlamp excitation source is selected by a 250-nm hand-pass interference having a 12% transmittance at 254 nm. The filtered light passes into a 10 mm by 10 mm by 45 mm quartz cuvette placed 10 mm from the interference filter. Fluorescence emission enters the spectrograph through the 250 pm wide hy 5 mm tall slits of the 200 grooves/mm, 200 mm focal length f / 3 concave holographic grating spectrograph (Model UFS-200, Instruments SA). The spectral coverage across the 320 columns of the CCD is approximately 240 nm. The 10-8 integration time is controlled hy an electromechanicalshutter. Approximately 2 mL of solution is used in each cuvette; however, on the basis of geometrical optics (17) the effective volume of sample from which fluorescence emission is collected is 1 mL. CCD Detector System. The two-dimensional m a y detector employed in these studies is a RCA-StDSOlEX 512 by 320 element CCD with 30 by 30 pm square detector elements. The electrooptical properties of the RCA family of CCDs is well characterized (16,1&21). This backside-illuminatedCCD was modified hy the manufacturer to enhance the ultranolet quantum efficiency by removing the protective glass surface covering the photoactive area and thinning the hulk silicon. The CCD is contained in a liquid nitrogen cooled Dewar and operated at 133 K to reduce the dark current. The support electronics for the CCD consists of a camera head and camera controller (Photometrics, Ltd., Tucson, AZ). The magnitudes of photogenerated charge packets are digitized to 14 bits and the digitized images are stored and processed hy a Motorola 68000 based computer (22). The CCD is operated in either a normal readout at high light levels or a 200-fold parallel binned readout mode at low light levels. Parallel binning of charge is the transfer and on-chip summation of photogeneratedcharge packets in the parallel shift direction along a column. Reagents. Reagent grade anthracene (98% pure, Aldrich Chemical Co.) was used without further purification. Ethanol (100%) was redistilled once in an all-glass distillation apparatus, with the first and last portions of the distillate discarded. Solutions helow M were used within several hours of dilution from stock solutions. The solutionswere not degassed. Glassware was cleaned in 50% nitric acid followed hy repeated rinsing with the redistilled ethanol. Procedure. The fluorescence and scattered light are focused onto the CCD with the slit dimension perpendicular to the serial

Orientation Figure 2. Orientation of anthracene spectrum on CCD. Wavelengh is dispersed in the serial charge transfer direction and the slit is focused along the parallel transfer direction.

ANALYTICAL CHEMISTRY, VOL. 61, NO. 3, FEBRUARY 1, 1989

284 25,000

O

r

220

260

340

300

380

420

460

Wavelength fnd

Flgure 3. Spectrum from lo-' M anthracene in ethanol for a 10-s integration. The anthracene fluorescence spectrum is centered about 400 nm. The elastically scattered light peak is at 254 nm and the Raman scattered light peak is at 274 nm.

detector drift occurring between the measurement of the blank and sample. This instrument utilizes the elastically scattered signal a t 254 nm or the Raman scattered light at 274 nm to correct for instrumental drift. The three intensities recorded for every image correspond to the photogenerated charge from the portion of the CCD illuminated by the fluorescence emission from the sample, Ifs, the elastically scattered excitation light, I,, and the Raman scattered light, I,. The intensities of the same three regions of the CCD are recorded for a 10-9 exposure of the blank to obtain the signals corresponding to the blank fluorescence, Im, blank elastically scattered, I&, and the blank Raman scattered intensities, Irk The fluorescence due to anthracene is equal to the sample fluorescence minus the blank fluorescence corrected for any drift between the two measurements. The drift correction is accomplished by normalizing the fluorescence signal by either the elastically or Raman scattered signal according to zelastically scattered corrected

IRaman scattered corrected

360

380

400

420

440

460

Wavelength ins)

Flgure 4. Spectra of 10" to lo-'' M anthracene for 10-s integrations. Spectra for normalized to constant intensity and offset for clarity.

register as shown in Figure 2. The spectrum consists of the four anthracene fluorescence peaks between 370 and 460 nm, the Raman scattered peak from ethanol at 274 nm and the elastically scattered peak at 254 nm. The total spectrum acquired across the CCD from a lo* M solution is shown in Figure 3. For each 10-8 exposure, the intensities from the three subarrays of the CCD corresponding to the regions of the fluorescence emission, the elastically scattered light, and the Raman scattered light are recorded. The blank and sample intensities are measured 10 times at each concentration to determine that average and standard deviation of the fluorescence signal. Binning of charge 200-fold in the parallel direction (the slit direction of the image) is used to increase the SIN ratio by reducing the total detector read noise contribution (13, 22). Binning in the parallel direction serves to compress the photogenerated charge in the slit dimension into a single 320-element row of the CCD. At high concentrations, the photogenerated charge is read in a normal readout mode to avoid oversaturating the charge holding capacity of the serial register. Figure 4 shows the unsmoothed anthracene fluorescence spectra from 10" to lo-" M solutions of anthracene. Each spectra is a 10-s integration of the ethanol blank subtracted from a 10-s integration of the anthracene solution. R E S U L T S AND DISCUSSION Analytical Calibration Curve. The\ sensitivity at low concentrations is determined by the ability of the+strument to accurately subtract the background signal of the blank from the background and fluorescence signal of the sample solution. At low concentrations the fluorescence signal is much smaller than the background signal. The accurate subtraction of the background requires correcting for any excitation source or

= Ifs/Ies - Ifb/Ieb

= Ifs/Irs - lfb/zrb

(2)

Although either eq 1or 2 can be used to correct for source or detector drift at low concentrations, slightly better results were obtained by using the elastically scattered peak. The intensity of the elastically scattered peak was 10 times the Raman scattered peak. The correction is most significant at low concentrations when the background fluorescence is equal to or greater than the fluorescence emission due to the analyte molecules. The working curve is linear from approximately lo4 to M. At concentrations above lo4 M, the instrument response drops off due to reabsorption of the short wavelength fluorescence and poor collection of fluorescence emission from the wall of the cuvette near the source. The linear correlation coefficient of the log-log curve in the to M concentration range is 0.99993 with a slope of 1.014. I t is important to note that the CCD easily measured fluorescence intensities ranging over 6 orders of magnitude without changing the integration time, source intensity, or optical throughout. Limit of Detection. The concentration limit of detection (LOD) at a S I N ratio of 2 for a 10-s integration determined by using the background correction described by eq 1 is 1 X M (227 parts per quadrillion). Taking into account the volume of solution effectively sampled by the entrance slit and grating, the LOD corresponds to 1 fmol, or approximately 6 X lo8 molecules. This LOD is some 300 times lower than work done employing a 60-W deuterium lamp and a P M T operated in a photon counting mode (23) and approximately 25 times longer than work done employing a tunable dye laser excitation source and a gated PMT (2). Without the correction technique described in eq 1 or 2, the LOD is lo-" M. The 10-s integrated signal due fluorescence emission of M anthracene is 2800 photogenerated e- superimposed on the background intensity of 260 000 photogenerated e-. The dominant source of noise at the detection limit is the uncertainty in the background intensity. The standard deviation in the background intensity is 1400 e-, or 0.5% of the background intensity. Although the main purpose of this work was to evaluate the quantitative capabilities of the fluorometer, it should be noted that the CCD did not achieve its sensitivity at the expense of spectral resolution as noted by the unsmoothed spectra of anthracene at lo-" M in Figure 4. CONCLUSIONS The results of this work demonstrate the improvements in sensitivity and dynamic range possible through the use of a low noise, two-dimensional CCD array detector. This point is emphasized by the use of a relatively weak and inexpensive

Anal. Chem. 1989, 61 285-288 I

excitation source. In addition to a gain in sensitivity, the instrument also acquires the complete molecular spectrum thereby increasing the qualitative information obtained with every measurement. The ability to normalize the fluorescence signal to the scattered light signal pro6des a simpler and more direct method of correcting for source and readout detector drift than methods employing beam splitters and auxiliary detectors. The results of this work point to the use of the CCD fluorescence spectrograph as a practical detector for use in high-performance liquid chromatography and flow-injection analysis.

ACKNOWLEDGMENT The authors thank Gary Sims and Richard Aikens of Photometrics, Ltd., and Robert W. Fitts of RCA for their assistance and support in the design and use of scientific grade CCD detector systems.

LITERATURE CITED Bradley, A. B.; Zare, R. N. J. Am. Chem. Soc. 1976, 98, 620. Rlchardson, J. H.; Ando. M. E. Anal. Chem. 1977. 4 9 , 955-959. Roach, M. C.; Harmony, M. D. Anal. Chem. 1987, 5 9 , 411-415. Talml, Y.;Busch, K. W. Guideilnes for the Selectlon of Four Optoelectronic Image Detectors. I n Multlchsnnel Image Defectors; Talmi, Y., Ed.; ACS Symposium Serbs 236; Amerlcan Chemical Society: Wash Ington, DC, 1983; Voi. 2. Chapter 1. (5) Jones, D. G. Anal. Chem. 1985, 5 7 , 1057A-1073A. (6) Jones, D. 0. Anal. Chem. 1985, 5 7 , 1207A-1214A. (7) Ryan, M. A.; Mlller. R. J.; Ingle, J. D. Anal. Chem. 1978, 5 0 , 1772-1 777.

(1) (2) (3) (4)

285

(8)Jadamec, J. R.; Saner, W. A.; Talml, Y. Anal. Chem. 1977, 49, 1318. (9) Johnson, D. W.; Callis, J. B.; Christian, G. D. Anal. Chem. 1977, 49, 747A-757A. (10) Walden, G. L.; Winefordner, J. D. Spectrosc. Lett. 1980, 13, 705-792. (11) Gluckman, J. C.; Shelly, D. C.; Novotny, M. V. Anal. Chem. 1985. 5 7 , 1546-1 552. (12) Hofstraat. J. W.; Engelsma, M.; de Roo, J. H.; Gooijer, C.; Velthorst, N. H. Appl. Specfrosc. 1987, 4 1 , 625. (13) Denton, M. B.; Lewis, H. A.; Sims, G. R. Charge-Injection and ChargeCoupled Devices in Practical Chemical Analysis. I n MuMhsnnel I m age Defectors; Talml. Y., Ed.; ACS Symposium Series 236; American Chemical Society: Washington, DC, 1983; Vol. 2, pp 133-154. (14) Sweedler, J. V.; Bllhorn, R. B.; Epperson. P. M.; Sims, G. R.; Denton, M. B. Anal. Chem. 1988, 60. 282A-291A. (15) Epperson, P. M.; Sweedler, J. V.; Biihorn, R. B.; Sims. G. R.; Denton, M. B. Anal. Chem. 1988, 60, 327A-335A. (16) Fowler, A.; Waddel, P.; Mortara, L. I n SolM State Imagers for Astronomy; Geary, J. C., Latham, D. W., Eds. Proc. SPIE-Inf. SOC.Opt. Eng. 1981, 290, 34-44. (17) Nlelsen, J. R. J. Opt. SOC. Am. 1930, 2 0 , 701-718. (18) Geary, J. C.; Kent, S. M. I n SolM State Imagers for Astronomy, Geary, J. C., Latham, D. W., Eds. Proc. SPIE-Int. SOC. Opt. Eng. 1981, 290, 51-57. (19) Waddel, P.; Christian, C. Opt. Eng. 1987, 2 6 , 734-741. (20) Leach, R. W. Opt. Eng. 1987, 26, 1061-1066. (21) Marien, K.; Pitz, E. Opt. Eng. 1987, 2 6 , 742-746. (22) Epperson, P. M. Ph.0. Dissertation, Department of Chemistry, Univershy of Arizona, Tucson, AZ, 1987. (23) Schwarz, F. P.; Wasik, S. P. Anal. Chem. 1976, 46, 524.

RECEIVED for review February 16, 1988. Resubmitted November 1,1988. Accepted November 4,1988. Portions of this work were supported by the Office of Naval Research.

Hydride Generator System for a 1-kW Inductively Coupled Plasma J. David Hwang,* Gary D. Guenther, and John P. Diomiguardi Occidental Chemical Corporation, Technology Center, 2801 Long Road, Grand Island, New York 14072

INTRODUCTION There are several valuable analytical characteristics that make inductively coupled plasma atomic emission spectroscopy (ICP-AES) a widely utilized atomic emission technique for single and multielement major, minor, and trace analysis. These include low detection limits for a large number of elements, calibrations that are linear over a very wide range (generally 5 to 6 orders of magnitude), low matrix or chemical interference effects, fast operation, and multielement determination capability (1-8). However, some hydride-forming elements such as arsenic, antimony, bismuth, selenium, tin, etc., are often present in real-world samples (e.g. environmental samples) at concentration levels too low to detect or determine by ICP-AES when conventional sample introduction techniques or ultrasonic nebulization are employed. Since each chemical form of a given hydride-forming element has its own characteristic environmental distribution and interactive effects on living organisms, the determination of ultratrace concentrations of hydride-forming elements in biomedical, clinical, toxicological, and environmental analysis has become an important topic. This has stimulated interest in the development of new analytical methods, or the adaptation of existing methods, for the determination of hydride-forming elements. The most commonly used technique for the introduction of the sample into the ICP is via a liquid aerosol generated by a pneumatic nebulizer. With the conventional nebulization technique, 99% of the sample solution is discarded (9, 10). The hydride generation technique offers several significant advantages over liquid aerosol introduction. For example, the transport efficiency of gaseous hydride approaches 100%. Hydride generation can be used as a preconcentration method,

resulting in improved detection limits. Gaseous hydride can also be separated from the sample solution matrix to virtually eliminate spectral interferences. Additionally, hydride sample introduction promotes more efficient plasma atomization and excitation. The desolvation and vaporization processes that occur in the plasma when aerosols are introduced are not necessary for gaseous hydride sample introduction. Therefore, the plasma energy normally required for desolvation and vaporization is available for atomization and excitation of the analyte, thus resulting in enhanced emission signals and improved sensitivity and detection limits. The historic development, basic properties, limitations, and recent applications of hydride generation techniques have been reviewed in some detail (10-14). One problem inherent to hydride generation that has delayed successful application of this method to the commerical ICPs is that the gaseous byproducts of hydride reaction, (H2, HzO, and C02) can extinguish a medium- or low-power ICP source when using the continuous-flow mode hydride generation approach. In order to keep the ICP from extinguishing, Thompson et al. (15-17') reported use of the hydride generation technique employing relatively high power (2.7 kW) ICP for the determination of arsenic, bismuth, antimony, selenium, and tellurium. Fry et al. (18) used a batch generation system that employed separation and low-temperature (liquid argon) condensation to remove undesirable reaction byproducts for the determination of arsenic with the use of 1.2 kW power ICP. In the experiment described below, a simple continuous flow mode hydride generation system has been developed for a 1-kW ICP-AES that has the advantages of both approaches described previously. Detection limits obtained by this system (As, 0.1 ng/mL; Bi, 1.4 ng/mL; Sb, 0.6 ng/mL; Se, 0.3 ng/mL;

0003-2700/89/0361-0285$01.50/00 1989 American Chemical Society