Anal. Chem. 1989, 6 7 , 593-600 (9) Van den Driest, Paul J.; Ritchie, Harald J.; Rose, Stephan LC-GC 1888, 6, 124-132. (10) Scott, R. P. W. J . ChrOmatOgr. S d . 1880, 18, 297-306. (11) Poole, Colin F.; Schuette, Sheila A. Contemporary Practice of Chromatography; Elsevier: Amsterdam, 1984; pp 220-222. (12) Papp, E.; Vigh, Gy. J . Chromatogr. $883, 259, 49-58. (13) Bij, Klaas E.; Horvath, Csaba; Melander, Wayne R.; Nahum, Avi J .
593
Chromatogr. 1981, 203, 65-84.
RECEIVED for review August 15, 1988. Accepted December 6, 1988. Partial financial support for this research was provided by NSF Grant CHE85-02061.
Design and Evaluation of a Dual Multichannel Detector Spectrometer for Simultaneous Molecular Absorption and Luminescence Measurements M. Cecilia Yappert,' M. W. Schuyler, and J. D. Ingle, Jr.* Department of Chemistry, Oregon State University, Gilbert Hall 153, Coruallis, Oregon 97331 -4003
A spectrometer that measures molecular absorbance (ABS) and fluorescence (FL) or chemiluminescence (CL) simultaneously Is described. The lower, mlddle, and upper portions of the test sdutlon are probed to obtain ABS, FL, and CL data. Optical fibers are used extensively. The ABS and FL or CL spectra are acquired in 400 ms or longer with diode array detectors mounted to spectrographs. The FL and CL signals can also be detected with photomultiplier tubes after wavelength seiectlon with fliters and/or a monochromator. Two mlcrocomputers in a master-slave arrangement control data acquisition. Useful absorbance data are obtained over the 300-800-nm range. The three dlfferent detection schemes are compared and the fluorescence detectlon Hmlts for quinine sulfate range from 1.2 to 40 pg/mL.
In studies of chemical equilibria and kinetics and for the analytical determination of molecular species, there are situations in which it is advantageous to obtain both luminescence [fluorescence (FL) or chemiluminescence (CL)] and absorbance (ABS) information about the same sample. Often only one of the species (a reactant, product, or intermediate) involved in a chemical reaction exhibits appreciable luminescence. By monitoring the luminescence changes due to the consumption or formation of this species and by following the other species spectrometrically, one obtains a more complete picture of the progress and changes that occur during a reaction. I t is also important to obtain absorption information when luminescence measurements are made under conditions where the concentrations of the analyte and/or other chromophores in the sample solution are large enough to cause inner-filter effects [significant attenuation of the excitation beam (primary absorption effect) or of the emission beam (secondary absorption effect)]. Only the latter effect is important in CL measurements. Significant analyte absorption causes nonlinearity in luminescence calibration curves. Absorption by concomitant species reduces the observed analyte luminescence signal and can cause error in the determination of the analyte (1-3). Clearly ABS, FL, and CL signals or spectra for a given solution or reaction mixture can be measured on three separate samples in three separate spectrometers. The ability to make Present address: Louisville, Louisville,
Department of Chemistry, University of
KY 40292.
all three types of measurements in one spectrometer from exactly the same test solution not only saves time but also enhances the precision of the analysis. Furthermore, in kinetics studies involving time-dependent information, the comparison and interpretation of the different types of spectral data are more reliable and straightforward. Several spectrometers have been described in the literature which are designed to make fluorescence and absorption measurements on the same solution (3-7). Typically, scanning monochromators with photomultiplier tube (PMT) detection have been employed. These spectrometers were used to generate absorption-corrected fluorescence signals or spectra that compensate for inner-filter effects. In this paper, we describe the design and characterization of a microcomputer-controlled spectrometer based on two solid-state, multichannel detectors (photodiode arrays) that allows simultaneous acquisition of FL or CL spectra and ABS spectra from the same solution. Sample throughput is enhanced with multichannel detection because spectral data are acquired in much less time (e.g., as little as 400~1s)than is possible with conventional scanning spectrometers. The multiple detector spectrometer also allows FL and CL emission to be monitored with PMTs. Detection limits are evaluated with P M T and multichannel modes of detection.
INSTRUMENTATION Optical Instrumentation. The design of the instrument is based on probing different volume elements in the sample for absorption (ABS), fluorescence (FL), and chemiluminescence (CL) signals as illustrated in Figure 1. A single source of radiation is used for both FL and ABS measurements and extensive use is made of optical fibers (0.f.) to reduce the number of optical components and alignment problems and to allow probing of different volumes of elements, in a standard 1 cm path length fluorescence sample cell, which are separated vertically by 1 cm. Diode arrays mounted to spectrographs are used for rapid acquisition of absorption and luminescence spectra. Filter/PMT and monochromator/PMT comrinations are also employed to measure luminescence signals. An automatic shutter is used to block the excitation beam to enable CL signals and spectra to be obtained from the center portion of the cell without interference from fluorescence. A diagram of the arrangement of optical components of the spectrometer is shown in Figure 2. Sources and characteristics of the components are listed in Table I. The monochroma-
0003-2700/89/0361-0593$01.50/0@ 1989 American Chemical Society
594
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
-i Figure 1. Sample cell configuration. Optical fiber 1 (o.f.1) directs white light to the cell; the transmitted radiation is collected by o.f.2 for the ABS measurement. Excitation radiation (AEx) is focused by a lens (not shown) into the central area of the cell 1 cm above the plane of o.f.l/o.f.2. Emission from the central portion is viewed at a right angle from two directions. Optical fiber 4 (o.f.4) collects luminescence in the plane of A,, (FL if the excitation beam is not blocked) from one direction. Emission from the opposite direction is also imaged by a lens (not shown) on the slit of a spectrograph/monochromator. Optical fiber 3 (o.f.3) is used to observe CL emitted from the upper pari of the sample cell
tors, spectrographs, and PMTs 2 and 3 are mounted to the outside walls of a light-tight aluminum box (the dashed line in Figure 2) which houses the sample cell and optical components. A 150-W Xe arc lamp is used as the source and provides both intense radiation in the 300-500-nm region for FL excitation and a relatively broad and reasonably flat spectral profile from 300 to 800 nm for absorption measurements. A elliptical reflector provides a high collection efficiency ( ~ 6 0 %for ) the source radiation. Fluorescence Measurements. The lamp beam is focused on a beam splitter (BSl) after passing through a water-based infrared filter (IRF). About 92% of the source radiation of interest is transmitted by BS1 and collected and focused by lens 1 ( L l ) on the entrance slit of the excitation monochromator. The lens position is chosen to just underfill the f number ( F / n ) of the monochromator. The excitation monochromator transmits part of the lamp beam over a selected wavelength interval. The beam leaving the monochromator exit slit (excitation beam) passes through a second shutter (S2) and is collected by a focusing lens (L3) and directed through another beam splitter (BS3). About 90% of the excitation beam is transmitted by BS3 and focused about 11 cm past the center of the sample cell. This provides a nearly collimated beam ( F / 1 0 ) to pass through the sample.
Within the sample cell, the beam width is about 3.5 mm when 2-mm excitation monochromatm slits are used. The reflected beam from BS3 is focused on a miniature vacuum phototube (PT). The signal from the PT is used to monitor and to compensate for drifts in the excitation beam intensity. The fluorescence from the central volume element of the sample cell is collected and collimated by L4 and then focused by L5 on the entrance slit of the emission spectrograph/ monochromator. Conditions were chosen to just underfill the F / n of the spectrograph/monochromator and the image magnification is 2. In the spectrograph mode, an intensified diode array (IDA) views the dispersed FL or CL radiation. The 1024-element IDA views a 512-nm wavelength range yielding a resolution of 0.5 nm per diode. To reduce the dark signal, the IDA is thermoelectrically cooled. A water-cooled collar installed on the heat exchanger allows the array to be cooled to about -10 "C (8,9). In the monochromator mode, a lateral mirror is installed inside the monochromator which directs the dispersed radiation to a variable exit slit for detection with PMT1. Fluorescence is also collected by 0.f. 4 which is juxtaposed to the opposite cell wall (see Figure 3). The emission radiation exiting the 0.f. is collected by a fast (F/0.75)aspheric lens (L7) and focused onto PMT3 after passing through a suitable emission filter. A similar combination of an optical fiber, lens, and P M T (0.f. 3, L6, and PMT2) is used to collect and detect CL radiation from the upper portion of the sample cell. Absorption Measurements. The reflected part of the source beam (approximately 8%)from BS1 is collected by and directed through a second beam splitter (BS2) by lens 2 (L2). About 10% of this beam is transmitted and focused on a reference photodiode (PD) detector after passing through a broad band UV filter (Fl, A, = 350 nm, 90-nm band-pass). The signal from this detector is used for optical feedback to stabilize the lamp intensity. The rest of the beam (=SO%) is reflected by BS2 and focused on an optical fiber bundle (o.f.1) after passing through a shutter (Sl). The beam entering o.f.1 is about F/1.9. Optical fiber bundle 1 directs the source radiation to the lower part of the cell. The end of o.f.1 is placed 5 mm away from a 1-mm-diameter aperture at the sample cell wall as shown in Figure 3. The resulting angular restriction changes the F / n of beam entering the sample cell from 1.9 to 5.0. This configuration ensures that all light rays used to measure the absorption travel approximately the same path length and minimizes radiation not used for absorption detection from entering the sample cell that could be reflected and interfere with luminescence measurements. The loss of radiation
I PMTI
F-1
Monochrom.
Flgure 2. Schematic diagram of the optical components. For reasons of clarity, O.f.4 is drawn to the right of o.f.3 (broken lines) even though it is actually positioned below o.f.3. Key: L, lens: BS, beam splitter; PD, photodiode; S, shutter; F, filter; o.f., optical fiber: C, sample cell; DA, diode array; IDA, intensified diode array.
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
595
Table I. Instrument Components item Xe arc lamp lamp housing power supply feedback system IR filter lenses 1 and 2 lenses 3 and 4 lens 5 lenses 6 and 7 beam splitter BS1 beam splitter BS2 beam splitter BS3 optical fiber o.f.1 optical fiber o.f.2 optical fiber o.f.3 optical fiber o.f.4 shutter 1, S1 driver unit for shutter 1 shutter 2, 52, and driver filter 1
reference vacuum phototube PMTs (1,2, 3) PMT power supplies PMT housing diode array (DA) system intensified diode array (IDA) system
model number/description Lamp System 901C-0011, 150 W, 20 V, 7.5 A ALH-220, water cooled, F/2.5, provided with socket extenders M302 450-W power supply with 302 starter TX-5, provided with photodiode sensor ALHl Optics biconvex, 25-mm diameter, 25-mm focal length, quartz biconvex, 25-mm diameter, 50-mm focal length, quartz biconvex, 25-mm diameter, 100-mm focal length, quartz aspheric lens, 24-mm diameter, 18-mm focal length, glass quartz, 25-mm diameter, l/le-in. thickness neutral density filter, 10% transmittance 25-mm diameter 0.2-mm thickness (microcover glass) UVP-1SllM400 MM1, UV light pipe: 1-mm bundle diameter, 100-pm fiber diameter, 40-cm length with stainless steel shielding UNP-lS2MlOOMM1, UV light pipe: 2-mm bundle diameter, 100-pm fiber diameter, 10-cm length with stainless steel shielding FLP-lSlB4S light pipe: 3-mm bundle diameter, 10-cm leneth. 56-um fiber diameter with stainless steel shielkng ' FLP-lSlBlOS, same as above except 25 cm long home-built Model T P 3.9X9 pull-type tubular solenoid, 24 V dc UNIBLITZ 214L4AOX5: programmable shutter, 14-mm aperture, SD122 driver unit band-pass filter: 350-nm maximum transmittance, 90-nm half-width Detector Systems R1842 1P28 227 3150s TN-1223-2 includes detector-amplifier head with 512 elements (50 pm X 450 pm) (Reticon, RL512 EC117) TN-1223-31 includes detector-amplifier head equipped with 1024 elements (25 pm X 2.5 mm) (Reticon RL 1024S),intensifier (ERMA photocathode), TN-1710-38 intensifier control module
source Canrad Hanovia, Newark, NJ Photochemical Res. Associates, Inc., ON, Canada same as above same as above same as above Rolyn Optics, Covina, CA Ealing, Newport Beach, CA same as above Melles-Griot, Irvine, CA Ealing, Newport Beach, CA same as above VWR Scientific, San Francisco, CA Welch Allyn, Inc., Skancateles Falls, NY same as above same as above same as above Newark, Chicago, IL Vincent Associates, Rochester, NY Rolyn Optics, Covina, CA
Hamamatsu, Hamamatsu City, Japan RCA Pacific Precision Instruments, Concord, CA same as above Tracor Northern, Middleton, WI same as above
Wavelength Selectors 1200-UV-H10 F/3.5, 100-mm focal length, concave Instruments, SA Inc., Metuchen, NJ holographic grating, 30 X 30 mm, 1200 grooves/mm. 250-nm blaze, 8 nm/mm dispersion, 1020-MA stepping motor and controller emission spectrograph/monochromator HR-320-260 F/4.8,58 X 58 mm ruled grating, 150 same as above grooves/mm, 500-nm blaze, 20 nm/mm dispersion, variable slit assembly (0-2 mm), lateral entrance mirror, HR-320-06 lateral exit mirror, HR-320-00 adaptor for Tracor IDARRT system, HR-320-25A stepping motor TN-1150 F/3 spectrograph concave holographic Tracor Northern, Middleton, WI ABS spectrograph grating, 25 nm/mm dispersion
excitation monochromator
sample cell stirbar magnetic stirrer injector
Sample Cell System Hellma Bel Art, F37 150 T-4656-00 submersible magnetic stirrer Model 77000 equipped with two electronically controlled 3-way pneumatic valves (Skinner Model MB 0002) and a solid state relav- (Gravhill model . . 7052-04-B-02-H)
throughput is not critical as absorption measurements are not light-limited. One end of optical fiber 2 (of.2) collects the transmitted radiation and directs it to the absorbance spectrograph. The other end of o.f.2 is held by a positioning device 4 mm away from the entrance slit. This distance yields the highest throughput without overfilling the spectrograph collimator. The transmitted light is dispersed and detected by a diode
Scientific Products, McGaw Park, IL same as above Cole-Parmer Instrument Co., Chicago, IL Hamilton Co., Reno, NV
array (DA). The 512-element DA is configured to view the 175-815 nm range with a resolution of 1.25 nm per diode. It is cooled by using the same method described for the IDA. The sample cell is masked on all four sides with appropriate circular apertures for all entering and exciting beams (see Figure 3). Provision is made for stirring the contents of the sample cell and for injection of a reagent to initiate a reaction with an automatic syringe injector. A set of baffles in the
596
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
(b) 0.f. 3
0.f. 4
Figure 3. Schematic drawing of optical fiber holders and sample cell: apertures (a) for ABS measurements and FL excitation and (b) for emission collection. Key: Left-leaning hatched lines, fiber bundle: dotted area, stainless steel ferrule; right-leaning hatched lines, 0.f. holder. Key: D , , 14 mm; D,, 23 mm; D,, 33 mm; A ,, A,, A,, A,, 3.5 mm. The holders for 0.f. 1 and -2 also serve as the sample cell holder.
sample compartment restricts scattered light from reaching the detectors. Microcomputers and Interfacing. To obtain absorption and luminescence spectral information simultaneously from both diode arrays, two microcomputers are operated in a master-slave network. Both microcomputers are AIM-65 (Rockwell International) microcomputers that were expanded with additional 1/0lines, RAM, floppy disks, a CRT terminal, a printer, and a H P 7470 dual pen digital plotter. The digital interfacing of the computers to various components is through the 1/0 lines of three 6522 1/0 chips of each computer system as shown in Figure 4. The 1/0 lines from two 6522 chips (one on each computer) are interconnected to provide a parallel bidirectional eight-bit data path for communication and synchronizing handshaking between the microcomputers. A second 1/0chip from each computer is interfaced to the IDA or the DA. The two control signals (the begin scan (BS) and clock (CK) signals) are supplied by two 1/0lines and are routed to the DA or IDA through a logic and timing (L/T) board that has been previously described (8, 9). The BS signals determine when a diode array is interrogated and are used to clear the diode array or to determine the exposure or integration time for signals. The diodes are sequentially interrogated at a rate determined by the CK frequency (38-ps period during data acquisition). During interrogaton of the diodes after an integration period, the maximum value of the voltage signal output on the diode array video line is sampled, digitized, and stored in computer memory with a sampleand-hold amplifier/analog-to-digital converter (SHA-ADC) which has previously been described ( 4 9 ) . The 12-bit ADC yields a resolution of about one part in 4000 for light intensity signals near saturation. Three 1/0 lines from the third 1/0 chip in the master microcomputer system control the reagent injector and the states of shutters 1 and 2. The third 1/0 chip in the slave microcomputer system is used for data acquisition from the
single-channel photodetectors. As seen in Figure 4, the anodes of each PMT and the reference PT are connected to separate current-to-voltage (I/V) converters with adjustable gain (i.e., feedback resistance) and time constant. The voltage outputs are mult.iplexed to a 12-bit ADC with a 25-1s conversion time. Software. The master system and software are used to (a) interact directly with the user, (b) acquire data from the IDA, (c) control the shutters and the injector, (d) calculate and present the data, and (e) communicate with the slave. Functions a and c are programmed in BASIC while tasks b, d, and e are performed in assembly language. The slave program, written in assembly language, permits data acquisition from the DA and three PMTs and the reference phototube. Only the general functions performed by the software are discussed here. Details of software and hardware are discussed in a thesis (10). The user via menus selects the tasks to be performed which can include acquisition of FL spectra from the IDA, CL spectra from the IDA, ABS spectra from the DA, FL P M T signals, and CL P M T signals. Next, the user is prompted to input data acquisition parameters, some of which are transferred to the slave system. At this point the sample (or blank) and reagent solutions are added to the sample cell. Once the computer is triggered, dark spectra and signals are acquired, a reagent is injected to initiate the desired analytical reaction, and then luminescence and ABS spectra and signals are acquired. If the ABS FL CL option is specified, the series of synchronized events shown in Figure 5a is implemented. The user inputs separate integration times for the FL and CL spectra, the number of FL/CL sequences (a sequence is a a sequential pair of FL and CL spectra separated by a user specified time), the time between sequences, and the time between the injection of the last reagent and the acquisition of the first spectrum in the sequence. The integration times for CL and FL spectra can be different to optimize performance so two dark spectra are taken by the master system corresponding to these integration times. Note that the dark spectra for the DA and IDA and dark photodetector signals are obtained before injection with shutters 1and 2 closed. The readings of the single-channel photodetectors are obtained by the slave immediately after each DA spectrum. Similar procedures and sequences are provided for the ABS CL FL task (Figure 5b), the ABS FL task, and the ABS + CL task. For the DA, which detects transmission spectra of the blank and sample solutions, a nominal integration time of 400 ms was used independent of the IDA integration time. This time was chosen to allow enough radiation to reach the photodiodes so that the signals for the diodes intercepting the spectral zone of largest intensity (around 470 nm) are near the saturation value. To ensure a constant base line for the diode arrays signals (i.e., complete erasure of previous signals), the diode arrays are refreshed (interrogated) 10 times before the first dark spectrum is taken and 5 more times before any subsequent spectrum is acquired (9). These refresh scans are conducted with a CK period of 6 ws without storage of data. To increase the total number of spectra that can be stored in the limited memory space, four adjacent diode signals from the IDA are summed into one data word location (2 bytes) to yield a 256-point luminescence spectrum with a resolution of 2 nm per data point. Likewise the signals from two adjacent elements in the DA array are summed into one data word to yield a 256-point transmission spectra with a resolution of 2.5 nm per data point. After the blank and sample measurements for a given task are completed, the DA and photodetector data acquired by
+
+
+
+
+
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
587
SLAVE
1 6 5 COMMUNICATION
2
I
\
2
#
CA I
0
CA2
2
Y
CA I
-
CA 2
1
# 0
l 2
5
I
I
I
F
d
I I
-
0
a
2 Injector
I, '
-
REF P T
--.C
I/V8
'-r
PMT I
-p
I/VI
.---t
PMT 2
-v
I/V2
---*
PMT3
--c
I/V3
---c
CB2 PA4
2
# 2
E
PA0 ? >
Figure 4. Components and connections of interfacing hardware. The asterlsk indicates that the interface circuitry for the DA is identical with that for the IDA. The ADC and multiplexer (MUX) are Analog Devices Models AD574 and AD7501, respectively. Key: M, mode control: S and SC, start conversion: EOC, end of conversion.
the slave system are transferred to the memory of the master system. At this point, a number of display and calculation options are available to the user and include: subtraction of dark or background spectra, calculation of transmittance and absorbance spectra, plotting of raw, corrected, transmission, or absorbance spectra, basic statistical treatment of the data, and correction of luminescence spectra and P M T signals for primary and secondary absorption effects based on the absorbance spectra. This last option is discussed in more detail in another paper (11). Diode array and P M T signals can also be compensated for source drift. For FL signals, the dark-corrected FL signal is ratioed to the reference PT signal. For ABS spectra, compensation for source intensity variations is based on the measurement of the lamp signal in a wavelength region where the sample does not absorb. The ratio of the mean signal in this selected wavelength region for the reference spectra to that for the sample transmission spectra is computed. The other data points in the sample transmission spectrum are multiplied by this ratio.
EXPERIMENTAL SECTION All solutions were prepared with double deionized water (denoted MW hereafter) obtained from a Millipore Milli-Q system fed by house-deionized water. All chemicals were reagent grade and the glassware was cleaned prior to use as previously described (12).
Most FL studies were conducted with quinine sulfate (QS) as the fluorophore. The excitation wavelength (&) was 359 nm and the excitation spectral band-pass (sex) was 16 nm. When the QS FL was monitored with the emission spectrograph/monochromator and the IDA or PMT1, the emission wavelength was always set to the wavelength of maximum FL, 461 nm, and these spectral band-pass was set to 20 nm, equivalent to 40 diodes or 10 data points for the IDA. When the QS FL was monitored with 0. f.4/PMT3, a low cut-off filter with a cut-off wavelength of 420 nm was employed. For ABS measurements, the entrance slit of the spectrographwas 150 bm. Thus,the resolution of ABS spectral information,limited by the slit width and the spectral band-pass, is 3.75 nm, equivalent to three diodes. For CL measurements with o.f.3/PMT2, either no filter was used or a low cutoff filter with a cut-off wavelength of 420 nm was employed. The performance of the instrument as a spectrofluorometer was evaluated by using test solutions with concentrations ranging from 0.1 pg/mL to 10 rg/mL of QS. These solutions were prepared by serial dilution with the blank (0.05 M H,SO,) in volumetric flasks. For some QS concentrations below 100 pg/mL, measurements were also performed on test solutions made by dilution in the sample cell. Here, a 10- to 1OO-gL aliquot of an appropriate test solution of QS was used to spike 2.5 mL of the blank solution in the cuvette in order to reach the desired final QS concentration. For studies involving CL measurements, the CL reactions of lucigenin (Lc) and luminol were employed. For studies of the CL of luminol with hypochlorite (13),2.5 mL of 3.2 X M luminol in a 0.05 M pH 10 borate buffer was introduced into the
598
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989 ( S H U T T E R I r e m a i n s open ) n
(a) OPEN SHUTTERS
CLOSE SHUTTERS F L DARK MASTER
(IDA)
I 1
1
INJECT
1
T2
4
T I .4 )
I
w
20
CL 2
T2
12
TRANS@
DARK2
-
T2
'r
L
SEQUENCE # 2
SEQUENCE % I
DARK1
CLOSE SHUTTER2
----/--I
I '
t
&*
TI
\
OPEN SHUTTER 2
I CL. I
FL. I
CL LARK
-
/ CLOSE SHUTTER2
TRANS1
TRANS2
TRANS3
TRANS4
SLAVE
(DA) ( S H U T T E R I remalns o p e n )
n
(b) CLOSE SHUTTERS
INJECT
C L DARK MASTER
-
(iDA)
/OPEN SHUTTER 2
T2
FLDARK
41
CL. I
TI
{A+
T2
-
.-. t i
CLOSE SHUTTER2
FL I
CL.2 i
TI
-_ -T
12
T 2 rn
'
OPEN SHUTTER2
FL.2
t ,
***
TI
c
is not taken in task b. cell and 1.0 mL of a 52.5 ng/mL OC1- solution was injected to start the reaction. For studies of the CL emitted during the reaction of Lc with HzOz (14), 1.0 mL of MW, 0.5 mL of a 0.1 M M Lc solution were H 2 0 2solution, and 1.0 mL of a 2.7 X introduced in that order into the cuvette with Eppendorf pipets. To start the reaction, a 0.5-mL aliquot of 5.0 M KOH solution was added with the automatic injector. For absorption measurements, QS was also the analyte. Measurements were made on test solutions over the 1.0 X to 10 pg/mL concentration range.
RESULTS AND DISCUSSION Design Evaluation. Because ABS and luminescence data are acquired simultaneously, it is important that the white light used for ABS measurements does not significantly affect the blank luminescence signals. If some of the white light output of o.f.1 is scattered and eventually collected and detected, equivalent increases in the blank and sample luminescence signals would result unless a significant portion of the scattered radiation was absorbed by species present only in the sample solution. Increased blank luminescence signals could increase the blank noise and degrade detection limits. The white light beam could also excite FL that can be scattered and collected by the emission optics. Blank signals, FL signals from QS, and CL signals from the Lc reaction were measured in the absence and presence of white light from o.f.1 (SIclosed and open). The blank noise and the net FL and CL signals are not affected by the white light. Therefore, the instrument allows ABS measurements to be made simultaneously with FL or CL measurements without a n y degradation in luminescence detection limits. To test the possibility of acquiring simultaneously FL signals (center volume) and CL (top volume) signals, the effect of FL on CL measurements was evaluated. The contribution of FL emitted from the central part of the cuvette to the radiation collected by o.f.3 from the upper volume element was first measured. It was determined that o.f.3 collects about 1% of the FL observed by o.f.4. With Lc and HzOz in the sample cell, the signal from o.f.3/PMT2 was monitored with and without the FL excitation beam (Aex = 366 nm) blocked
(S2 closed and opened, respectively) before and after the injection of KOH. The fluorescence signal in this case is due to Lc and one of the products of the reaction (i.e., 10methylacridone). No filter was placed before PMT2 because the CL and FL observed with the Lc system have similar emission maxima and cannot be spectrally resolved with fiiters. The net CL signal collected by o.f.3 is only 1.5% of the net FL detected by the same fiber bundle because the FL intensity is approximately a factor of 6.5 X lo3 greater than the CL intensity. Consequently, it is not possible to resolve FL and CL by spatial means when the intensity of the FL emission is much greater than that of the CL emission and when the CL and FL spectra overlap. To demonstrate the effect of intense CL on the FL measurements with o.f.4, the very efficient CL reaction of luminol with OC1- was used as the test system. Here the FL observed with excitation at 393 nm is due to the primary reaction product, 3-aminophthalate. The net FL signal detected with o.f.4 is only about 1% of the net CL signal and is difficult to distinguish. Because CL originates from the whole sample volume after mixing, o.f.3 collects the same amount of CL as collected by o.f.4. Therefore, in the absence of discrimination of FL from CL by wavelength selection, measurement of the FL signal is hindered if the CL intensity is comparable to, or greater than, the FL intensity. The results of the above studies indicate that the FL and CL signals cannot be spatially resolved to an adequate degree. To monitor the FL signal with o.f.4iPMT3 or the emission monochromator/PMTl configurations without significant interference from CL, the FL signal should be approximately a factor of 100 times greater than the CL signal. However, when the FL signal is about 100 times greater than the CL signal, the FL signal observed by o.f.3iPMT2 is approximately equal to the CL signal detected and the FL would interfere with the detection of CL. Consequently, it was decided to acquire CL and FL in a sequential, rather than simultaneous, manner by closing and opening S2. The sequential blocking of the FL excitation beam also allows the acquisition of CL spectra with the IDA without interference from FL. To
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
minimize the inevitable contribution of CL to the FL signals detected with PMT1, PMT3, or IDA, the excitation conditions (A,,, sex) are arranged so that the FL intensity is at least 100 times greater than the CL intensity. No detectable contribution of FL excitation radiation or FL (or CL) emission radiation to the radiation transmitted through the lower part of the cell (the ABS measurement) was observed. Absorption Spectrophotometer Performance. The spectrometer provides adequate spectrophotometric accuracy and precision over the 320-815 nm wavelength region. For wavelengths below 320 nm, the DA photodiode signals decrease rapidly due to the falloff in lamp radiance such that measurement uncertainty becomes unacceptable. The major problem observed with absorption measurements was stray radiation which limited the accuracy of the measurements a t high absorbance values. Measurements of transmission spectra of different cut-off and neutral density filters suggested that a significant fraction of the stray radiation originates from the reflection or scattering of the radiation dispersed on the quartz window covering the DA or from other optical components in the spectrograph. To reduce this problem which limited the maximum ABS values to 1.3 AU in the low wavelength region (near 350 nm), two solutions were implemented. First the 9.9 X 41 mm quartz window on the DA was partially covered with a black aperture so that only a 1.8 X 30 mm rectangular area in front of the diode array is exposed to radiation. Second, the spectral profile incident on the DA was changed to increase the relative contribution of radiation below 440 nm. This was accomplished by placing a contour mask inside the spectrograph. This mask is made out of black matte paper and is placed in the trajectory of the dispersed radiation 50 mm before it reaches the DA. The widest opening allows all of the low wavelength radiation to reach the DA. The narrowest part of the opening restricts the radiation detected at the 460-nm region to the greatest extent (i.e., about 45% transmission). The above modifications considerably reduce the amount of stray radiation and extend the ABS range for measurements a t 350 nm to above 2.0 AU. The percent stray radiation is quite wavelength dependent and still larger for lower wavelengths (typically about 0.5% at 350 nm). Nonlinearity above 1 AU due to stray radiation can still limit absolute accuracy. Therefore, a provision was made in the BASIC program used to calculate the ABS which allows the user to input the base-line intensity attributed to stray radiation. This value can be obtained by acquiring the transmission spectrum of a solution that has a very high absorbance in the wavelength range of interest. The program subtracts this input stray radiation value from the appropriate reference and sample signals before calculation of the transmittance and absorbance. Near the detection limit (DL), the noise in the blank measurements (standard deviation for a data point) varies from 0.3 to 0.6% of full scale, depending on the wavelength. This corresponds to a standard deviation in absorbance (uA) at A = 0 of 0.001-0.002 AU. At absorbances near 1.0, the relative standard deviation in absorbance is typically 0.5-1 % . Overall the performance of the spectrometer as a spectrophotometer is adequate but not as good as that obtained with high-quality conventional spectrophotometers using a scanning monochromator and P M T detection or commercial diode array spectrophotometers. Spectrofluorometer Performance. The FL of QS was monitored with the three possible FL collection/detection configurations, and the values of the calibration slope in the linear region, the theoretical detection limit (DL), and the theoretical blank equivalent concentration (BEC) are summarized in Table 11. The calibration curves for all detection configurations are similar in shape to those shown in Figure
599
Table 11. Performance Characteristics for the Three FL Detection Configurations collection/detection configuration
m: counts/ (pg/mL)
DL,c pg/mL
BEC," pg/mL
filter, o.f.4/PMT3 monochromator/PMTl spectrograph/IDA
13' 0.15
0.4 13 40e 2sr
100 27 220e 23d
5.1'
49
Slopes calculated from linear region of curve. Normalized to the PMT gain of monochromator/PMTl data. OTheoretical detection limit DL = 2sbk/m, sbk is the blank standard deviation. Time constant for PMTs, 1 s; IDA integration time, 5 s. dBlank equivalent concentration, concentration of QS yielding a net fluorescence signal equal to that for the blank solution. e Calculated for 1 data point (458-460 nm). /Calculated for 10 data points (447-467 nm). 9.0
I
3.0
1
to
tt
.
A
0.0
I
-1.0
,
I
i
1 1
I
0.0
1.0
1
2.0
3.0
4.0
1
5.0
8.0
7.0
&O
LOG OS CONCENTRATION @/mL)
Quinine sulfate fluorescence calibration curve obtained with emission monochromatorlPMT1 configuration. All signals are standardized to a feedback resistance of 10' fl and a PMT gain of 4.5 X lo6 (PMT bias voltage of 900 V). One count (from the ADC) is equivalent to 2.5 mV. (a)Solutions made by serial dilution in volumetric flasks; (b) dilution in sample cell. Figure 6.
6 for the emission monochromator/PMTl configuration. The range of linearity extends from about lo3 to lo6 pg/mL with all detection schemes. At lo7 pg/mL, there is a deviation of about 20% from linearity due to inner-filter effects. At concentrations below about lo3pg/mL (see curve a), the FL signals level off (i.e., they are greater than expected from extrapolation of the linear part of the calibration curve). This behavior has been observed previously and is attributed to equilibria involving the analyte a t low concentration and sorption/desorption of the analyte and contaminants not only in the flasks but also in the sample cell itself (15). By addition of small aliquots of the analyte to the blank solution in the sample cell, the contamination effect is diminished (see curve b). However, for lowest concentrations, the spiking procedure does not compensate for the contamination present in the low concentration standard solutions used for spiking. Contamination could also be introduced from the Eppendorf tips used for addition of spiking solution. The gains of all the PMTs were calibrated against a P M T whose gain was previously calibrated (16). This allows P M T signals to be normalized to one gain and comparison of the relative radiant power viewed with the two P M T detection configurations. From Table 11, the slope of the linear region of the normalized calibration curve obtained with the filter configuration is about a factor of 85 greater than that achieved
600
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
with the monochromator configuration. This occurs because the 420-nm cut-off filter allows the transmission of the complete FL emission profile, whereas with the monochromator, only a f 2 0 nm spectral region centered at the maximum of the FL band is transmitted. In addition, the product of the solid angle and volume element of solution excited and viewed with the filter/PMT configuration is about a factor of 3.5 greater than that obtained with the monochromator/PMT arrangement. The FL detection limit for QS obtained with a spectrograph and IDA detection is near to that obtained with a monochromator and P M T detector and is more than an order of magnitude better than that reported previously (17) for a multichannel detector. This improvement is due in part to the lamp housing which provides higher excitation radiant power and to the different IDA with taller individual diodes (2.5 vs 0.45 mm). The DL's reported in Table I1 are theoretical values evaluated from the slope in the linear region of the calibration curve and the blank standard deviation. Because of the nonlinearity at low concentrations, concentrations of QS below these DLs can be measured with good signal to noise ratio, and the practical and theoretical DLs differ (15). The filter/PMT configuration provides the best DL because the fluorescence flux observed is greater without a corresponding increase in the blank noise. At concentrations above IO5 pg/mL for all detection configurations, source flicker noise is limiting and the relative standard deviation in signals is typically 0.5%. The blank equivalent concentration is about a factor of 4 smaller with the monochromator configuration than that with the filter configuration. This is expected because the filter transmits radiation over a wider wavelength region than the monochromator where the ratio of the QS FL signal to the background FL signal is less. With the IDA, the BEC is the largest. Apparently, the stray radiation level is greater in the spectrograph mode because of scattering and reflection off the window of the intensifier.
CONCLUSIONS The multiple detector spectrometer developed is a unique and powerful tool for both analytical and fundamental studies. It performs the functions of three spectrometers at once. Luminescence (FL or CL) and ABS data are acquired from the sample solution at precisely controlled times so that the information provided by each technique can be readily and reliably compared. The rapid (on the time scale of seconds), simultaneous acquisition of absorption and luminescence spectra is crucial for kinetics studies or chemiluminescence
measurements in which concentrations of reactants, intermediates, and products are changing with time (18). Simultaneous measurement of the ABS at both the excitation and emission wavelengths allows the rapid correction of FL signals for primary and secondary absorption effects as is discussed in another paper (11). For quantitative luminescence measurements, the spectrometer performs as well as other high-quality FL or CL spectrometers, and it is unique in offering three different detection configurations. The best FL detection limits are realized with the o.f./filter/PMT detection scheme. The data demonstrate that ABS and FL measurements, or ABS and CL measurements can be made simultaneously without detrimental cross interferences. However, FL and CL measurements from the same test solution must be made sequentially by controlled blocking of the FL excitation beam. With the present system, the minimum data acquisition time for diode array spectra is 0.4 s. Hence it can be employed for reactions where the concentration of absorbing and luminescent species does not change significantly over a period of about 1 s.
LITERATURE CITED Ingle, J. D., Jr.; Crouch, S. R. Spectrochemical Anawsis: PrentlceHall: Englewood Cliffs, NJ. 1988; pp 449-458. Winefordner, J. D.; Schulman, S. G.; O'Haver, T. C. Luminescence Spectrometry in Analytical Chemistry; John Wiley and Sons: New York, 1972; pp 293-294. Christmann. D. R. Ph.D. Thesis, Michigan State University, 1980. Holland. J. F.; Teets, R . E.; Kelly, P. M.; Timnick, A. Anal. Chem. 1977, 49, 706-710. Holland, J. F.; Teets, R. E.; Timnick, A. Anal. Chem. 1973, 45, 145-153 . _. Christmann, D. R.; Crouch, S. R.; Holland, J. F.; Timnick, A. Anal. Chem. 1980. 52. 291-295. Ratzlaff, E. H:; Harfmann, R. G.; Crouch, S. R. Anal. Chem. 1984, 5 6 , 342-347. Ryan, M. A. Ph.D. Thesis, Oregon State university, 1980. Ryan-Hotchkiss, M.; Ingle, J. D., Jr. Tahnta 1987, 3 4 , 619-627. Yappert, M. C. Ph.D. Thesis, Oregon State University, 1985. Yappert, M. C.; Ingle, J. D., Jr. Appl. Spectrosc., in press. Wilson, R. L.; Ingle, J. D., Jr. Anal. Chem. 1977, 4 9 , 1060-1065. Marino, D. F.; Ingle, J. D., Jr. Anal. Chem. 1981, 5 3 , 455-458. Montano, L. A.; Ingle, J. D., Jr. Anal. Chem. 1979, 57, 919-925. Ingle, J. D., Jr.; Wilson, R. L. Anal. Chem. 1976, 4 8 , 1641-1642. Bower, N. W.; Ingle, J. D., Jr. Anal. Chem. 1975, 4 7 , 2069-2072. Ryan, M. A.; Miller, R. J.; Ingle, J. D., Jr. Anal. Chem. 1978, 5 0 , 1772-1777. Yappert, M. C.; Ingle, J. D., Jr. Appl. Spectrosc., in press.
RECEIVED for review May 2, 1988. Resubmitted November 23, 1988. Accepted December 5 , 1988. Acknowledgment is made to the NSF (Grant Nos. CHE-79-21293 and CHE-8401784) for partial support of this research. Presented in part at the 1983 Spring National ACS Meeting, Seattle, WA, and the 1987 Fall National ACS Meeting in New Orleans, LA.
