(14)S. A. Goldstein and J. P. Walters, Spectrochim. Acta, Part B, 31, 201 (1976). (15) S.A. Goldstein and J. P. Walters, Spectrochim. Acta, Part 8,31, 295, (1976). (16)W. J. Pearce in "Conference on Extremely High Temperatures", H. Fischer and L. C. Mansur, Ed., John Wiley and Sons, Inc., New York, 1958,p. 123. (17) . , C.E. Harvev. "SDectrochemical Procedures", ADDlied . . Research Laboratories, Gleidale,'Calif., 1950. (18)M. P. Freeman and S. Katz, J. Opt. SOC.Am., 50, 826 (1960). (19)P. J. Dickerman and R. W. Deuel, Rev. Sci. Instrum., 35, 978 (1984). (20) L. A. Luizova, Opt. Spectrosc., 38, 362 (1975). (21)J. P. Walters and S. A. Goldstein, How and What to Sample in the Analyticai Gap", in "Sampling, Standards, and Homogeneity", ASTM STP 540, 71 (1973).
(22)V. M. Kock and J. Richter, Ann. Phys., 24, 30 (1969). (23)D. W. Blair, J. Quant. Spectrosc. Radiat. Transfer, 14, 325 (1974). (24)w. L. Barr, J. Opt. SOC. Am., 52, 885 (1962).
RECEIVEDfor review January 28, 1976. Accepted June 18, 1976. The financial support of the National Science Foundation in the form of a traineeship (AS) and for computing (grant number GP-35602X) is acknowledged and appreciated. portions of this work were reported at the 2d annual FACSS meeting, Indianapolis, Ind., on October 7, 1975.
Simultaneous, Split-Beam, Ratio Measurement System J. D. Defreese' and H. V. Malmstadt* School of Chemical Sciences, University of Illinois at Urbana-Champaign, Urbana, 111. 6 180 1
A dual beam, dual detector, ratio measurement system which compensates for source fluctuations over a wide frequency range is described. Automatic correction for source flicker of high intensity as well as of more conventional sources and for pulse-to-pulse variations of short-duration (-10 ns) pulsed lasers is obtained by simultaneously acquiring source information and encoded chemical information. Thls is required if the full advantage of high photon fluxes for improved measurement precisionand sensitivity is to be realized. The ratio values typically show an order-of-magnitude or better improvement in precision compared to single-channel measurements. The system performance is also compared to optical feedback approachesfor source stabilization.
High intensity light sources are finding increased applicability for spectrochemical analyses (1, Z), particularly in kinetic methods (3, 4 ) where high precision is required for measurement of small changes in absorbance and in atomic fluorescence ( 5 6 )and molecular fluorescencedeterminations (7,8) where high source intensity leads to increased sensitivity. Photon statistical variations (9), which establish a fundamental limit on the attainable photometric precision, are improved by the high photon fluxes obtained from intense light sources. However, intense light sources in general are more unstable than conventional sources because power supply requirements make output regulation difficult. Source instability can also manifest itself as pulse-to-pulse variations in the output intensity of pulsed or programmed light sources such as lasers or electronically modulated continuum sources. To realize the full advantages of increased light intensity, Le., improved sensitivity and measurement precision because of higher photon flux, compensation for source flicker or pulseto-pulse variations must be made. For maximum precision, compensation may be important for even conventional, well-regulated spectroscopic sources. Compensation for both low and high frequency source noise can be obtained by employing double beam, time-shared, single detector systems which alternately sample the source radiation in reference and sample channels. However, these systems cannot discriminate effectively against noise whose frequency spectrum lies near the modulation frequency of the reference and sample channels. Also, when the signal of inPresent address, Department of Chemistry, University of Kansas, Lawrence, K a n . 66045.
1530
terest is present for too short a time to be sampled by both beams, as would be the case for short-duration laser pulses, source fluctuation correction is not accomplished, and, when measuring fast reaction rates, valuable sample information is lost during the time that source information is being obtained in a reference channel. Many of the limitations of the double beam, single detector systems can be avoided by using double beam, dual detector systems which acquire reference and sample channel information simultaneously. One type of dual detector system used for stabilizing source output is "optical feedback" (10-12) which has been shown to effectively stabilize light sourceswith fairly long time constants, most notably tungsten lamps (IO, 11, 13, 14). However, a fundamental limitation of optical feedback is that the correction being applied to the source at a given time is derived from information on a state of the source which existed at an earlier point in time. Consequently, attempts to stabilize sources with high frequency noise components such as mercury arc lamps have not been very successful (15). The optical feedback approach is also not applicable to short-duration pulsed light sources because the information is present for too short a time to provide meaningful feedback to the source. A double beam, dual detector, ratio measurement system which compensates for source flicker over a wide frequency range by simultaneously taking source information and encoded chemical information is described in this paper. The system is shown to provide automatic correction for source flicker of high intensity sources and conventional sources and for pulse-to-pulse variations of short-duration pulsed lasers. An evaluation of performance compared to an optical feedback approach is also presented.
INSTRUMENTATION A block diagram of the spectrometer system is shown in Figure 1 in a configuration for making absorbance measurements. In the simultaneous ratiometric mode of operation, the processor ratios the sample and reference beam information after automatically correcting for any background from detector dark current, charge-to-count converter offsets, etc. If desired, this ratio is automatically multiplied by a correction factor calculated during a reference cycle to produce a transmittance readout. For molecular fluorescence measurements, the configuration is changed from that shown in Figure 1. The beamsplitter module is placed before a fluorescence cell module which allows either 30" front face or 90" fluorescence to be gathered
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1 1 , SEPTEMBER 1976
’
feedback (OF) or constant voltage (CV) mode of operation could easily be switch-selected, I I C b I I READOUT I In the constant voltage mode, the voltage across the lamp COUNT U is maintained equal to an adjustable reference voltage (regVOLTAGE 1 DIGITAL CORRELATION ulated 15 V across voltage divider) by operational amplifier i n OAl wired as a voltage follower. The output of OAl is current boosted by transistors Q1, Q2, and Q3 wired as emitter followers. The zener diode a t the non-inverting input of OAl protects the lamp by limiting the voltage across it to 7 V. When in the optical feedback mode, the reference voltage CHARGE to OAl is obtained from OA:! and is inversely proportional to the output intensity of the tungsten lamp. Therefore, the SOURCE WAVELENGTH BEAM DETECTOR voltage across the lamp is varied in such a manner as to comI S O L A T I O N SPLITTER pensate for changes in output radiance of the lamp. The count Figure 1. Block diagram of the spectrometer system rate from the reference channel q-to-N is converted to a proportional voltage by R2 and C2 which form a pulse averager. For a full scale output from the q-to-N (50%duty cycle), the resultant voltage at VI will be approximately 2.5 V. The and redirected along the normal optical axis. The monovoltage applied to the lamp will equal the output voltage of chromator isolates the desired wavelength for the sample OA2 since, in the OF mode, it is the reference voltage for the detector. In this mode of operation, the background-corrected follower OAl. OA2’s output will vary in such a way as to cause ratio is the value of interest and no transmittance correction factor is used as in absorbance measurements. the voltages at its inputs to be equal. Therefore, the reference voltage a t the (+) input of OA2 sets the reference channel Optional “optical feedback” circuitry is shown within the frequency or, equivalently, the current output of the reference dashed lines of Figure 1. Although not normally used, this detector. For example, if the reference voltage is 1.25 V and mode of operation is included for comparison to the simultaneous ratiometric mode for correction of fluctuations of a the q-to-N converter is set on the 10-6-A and 1-MHz scales, the voltage on the lamp will be adjusted such that the current tungsten lamp source. In order to allow simple switch selection from the reference detector will be 5 X 10-7 A and the output of the mode of regulation, i.e., optical feedback to the source frequency will be 500 KHz, corresponding to 1.25 V a t VI. or constant voltage on the source, and to allow measurements Now, if the output of the light source increases, thz current to be made on the output of the reference channel detector when in an optical feedback mode, the “extra” data domain output of the reference detector and therefore the frequency conversions of q-to-N, N-to-V, and V-to-i were included in from the q-to-N converter will also increase, as will the voltage the optical feedback system rather than connecting the refa t VI. Since the voltage a t the (-) input is now greater than erence detector output directly to the summing point of the that a t the (+) input, the output of OA2 will decrease. Confeedback amplifier. These data domain conversions have sequently, the voltage across the lamp will be decreased (benegligible effect on the response time and accuracy of the cause of the increased voltage drop across Q1) and the current feedback system. from the reference detector will be readjusted to its original value as the radiance of the lamp drops. The value of capacitor Sources. Various light sources were used to obtain the results reported in this paper. Data on the long-term drift of the C1 is chosen to prevent the output of OA2 from changing more rapidly than the lamp filament can follow, thereby preventing system and data comparing the “optical feedback” and oscillation of the optical feedback circuit. “constant voltage” modes of operation were obtained using a CM1630 tungsten lamp (Chicago Miniature, Chicago, Ill.) A 420-W GE FAL tungsten halogen projection lamp is used rated a t 6.5 V and 2.75 A. The power supply circuit shown in to illustrate system performance with a continuous, high intensity source exhibiting large ripple. The ripple is obtained Figure 2 was constructed in such a way that either an optical OPTICAL FEEDBACK
FEEDBACK AMPLIFIER
I
I
i
REF q-to-N
PULSE - AVERAGER
5 v
OA2
CI
I5 v
6 REF
q-to-N
SET REFERENCE CHANNEL CURRENT
Figure 2. Lamp power supply
S E T LAMP VOLTAGE
CM 1630
and optical feedback circuitry
T
Capacitance values in fiF and resistance values in ohms unless otherwise noted. (a) Block diagram, (b) detailed schematic
ANALYTICAL CHEMISTRY, VOL. 48, NO. 11, SEPTEMBER 1976
1531
by operating the lamp directly off the ac line. The lamp is mounted in a modified Heath EU-701-90x tungsten light source module which has a cooling fan attached. This unit is then fastened directly to the monochromator using 2-inch standoffs to minimize heat transfer to the monochromator. The data presented for the programmed hollow cathode lamp were obtained with a Hamamatsu deuterium hollow cathode lamp (Model L233-1DQ).The D2 lamp is powered by a programmable, digitally-controlled, current-regulated supply (16). Modifications to this power supply included the use of an 800-V raw supply obtained by wiring two Heath EUW-15 universal power supplies in series and the replacement of the DTS 413 pass transistor with a higher voltage SK 3115. The higher voltage is required because the deuterium hollow cathode lamp has a higher ignition voltage than most elemental lamps. A nitrogen laser, and a nitrogen pumped dye laser were used in this study to demonstrate that the ratiometric system can compensate for variations in short-duration light pulses. The Nz laser has been designed and built in our laboratory (7).The laser tube, TTL compatible triggering circuit, and power supply are enclosed in a single aluminum plate box to minimize rf noise which presented a problem with a commercial system that was evaluated, apparently due to the long cables between the power supply and the laser module. The laser is operated with a dynamic nitrogen pressure and variable pulse rate up to 100 Hz. The dye laser is constructed by placing a cell containing an organic dye selected for a given wavelength region in an optical cavity formed either by a grating or by a reflecting mirror and a partially transmitting output mirror. The dye is transversely pumped at 337.1 nm by a superradiant Nz discharge (17).The output is tuned by setting the grating angle such that the first-order wavelengthsare directed back along the optical axis of the cavity. When the grating is perpendicular to the dye laser optical axis (zero order), broadband lasing is obtained. Broadband lasing can also be obtained simply by allowingthe parallel walls of the dye cell to form the laser cavity. Wavelength Isolation. A GCA/McPherson EU-700 monochromator is used in the spectrometer system for wavelength selection. Beamsplitters. The desirable characteristics of a beamsplitter are mechanical and thermal stability, a minimal variation in the reflectance/transmission ratio across the surface and with wavelength,and a wide wavelength coverage. The most commonly used beamsplitter is a 1-inch diameter, %6 in. thick, Suprasil quartz disk (Amersil,Inc., Hillside, N.J.) which reflects approximately 15% of the incident radiation. The quartz plate exhibits good mechanical and thermal stability and can be used throughout the uv and visible regions. A 1-inch glass cube (Edmund Scientific, Barrington, N.J.), a quartz cube (Esco Products, Oak Ridge, N.J.) and pellicles (National Photocolor Corporation, South Norwalk, Conn.) have also been employed as beamsplitters. Detectors. Either phototubes or photomultiplier tubes are used as radiation detectors in the measurement system. The reference and sample detectors are normally powered by a common supply to increase stability. The 1P39 phototubes (Hamamatsu Corp.) are operated at 125 V supplied by a Heath EU-4OA solid state high voltage power supply. The photomultiplier (PM) tubes are RCA 1P28A’spowered by a Heath EU-42A high voltage power supply. Each dynode resistor chain is modified for use in pulsed applications to avoid nonlinear responses of the photomultipliers (18).T o reduce space charge effects due to the large number of electrons generated in the area of the last few dynodes, the voltages from the ninth dynode to the anode (Rlo = 220 kO) and from the eighth dynode to the ninth dynode (Rg = 150 kQ) are larger than the remaining interdynode voltages (Rz-Rs = 100 kO). 1532
The voltage between the photocathode and first dynode is also made larger (R1 = 220 kO) to increase the collection efficiency of the first dynode. Capacitors from the last four dynodes to ground with values of 0.001, 0.002,0.005, and 0.01 pF for dynode six through dynode nine, respectively, provide increased current capability for the large current surges generated in the last few dynodes by intense light pulses. These capacitors prevent the dynode voltages from changing, thereby maintaining the gain of the photomultiplier tube constant. It became apparent as the spectrometer system was characterized that the movement of the position of impingement of the radiation on the nonuniformly sensitive photocathodes of the detectors was causing problems in terms of drift and ratio inconsistency for widely varying light levels. To help alleviate this positional sensitivity, the envelopes of the detectors over the photocathodes were frosted by sand blasting. In this way, the radiation is spread out over a wider area of the photocathodes and the effects of positional shifts are averaged out. The dispersion of the incident radiation may also increase the linear dynamic range of the detector in situations where localized saturation of the photocathode would otherwise occur. For the data reported here, the light loss due to scattering by the frosted areas presented no problem. q-to-N Converters. The charge-to-count converter (19) is an integrating analog-to-digital converter based on a servo charge comparison technique and has a sensitivity of A and 5 MHz full scale, a precision of better than 0.00196, and nonlinearity of less than 0.005%. The ability of the chargeto-count converter to integrate large charge pulses and still produce a proportional output count is a major factor in the applicability of the simultaneous ratiometric system to short-duration pulsed light sources. Data Acquisition and Correlation. A high speed Decimal Data Processor (DDP) (20, 21) is used for data acquisition, reduction, and storage. The required calculations on the spectrometer signals can also be performed by an inexpensive digital correlator (22) but the Decimal Data Processor is much more versatile because it is software programmed. The 8-digit decimal processor responds to a basic set of 11 instructions which cause it to operate on numbers stored in two registers, the input/output (I/O) and accumulator (AC) registers. Numbers are entered and retrieved via the 1/0register. A schematic diagram of the DDP and measurement interface is shown in Figure 3. The time base is obtained from a Heath EU-805 Universal Digital Instrument (UDI) which also visually displays the integrated output of the reference channel q-to-N converter. The output of the sample channel q-to-N is monitored on a second UDI. The monostables (MSI, MSII) are 74121 I.C.’s with pulse durations of about 1ps. Quad decade counting unit (DCU) cards (Heath Co., Benton Harbor, Mich., EU-800-DD) are used for the counters and latches which simultaneously integrate the counts from the sample and reference channel q-to-N’s and then latch this information. The 8-digit I/O register of the DDP is connected in parallel with the two 4-digit DCU cards, forming the interface between the processor and measurement circuitry. For synchronization when using a laser source, the laser trigger circuit simply converts the logic level time base signal to one compatible with the trigger input of the N2 pump of the dye laser which requires a positive 20-V pulse. For the N2 laser, the time base is connected directly to the trigger input. The waveforms for the measurement system are shown in Figure 4. On each negative-going edge of the time base, the gates to the counters are closed and the integrated counts are latched. Then, the counters are cleared and the cycle begins again. Therefore, the integration period for each measurement is equal to the period of the time base minus about 2 pa, 1ps that the gate control is low while the data are latched plus 1 I.LS during which the counters are cleared. The data from both
ANALYTICAL CHEMISTRY, VOL. 48, NO. 11, SEPTEMBER 1976
TELETYPEWRITER
T I M E BASE (UDII
I PROGRAM STORAGE
CONTROL
I
ARITHMETIC bND REGISTERS
STORAOE
I
CHART STRIP RECORDER
I
I 1
SbMPLE q-le-N
9F k m*m clr
1 -
KEVBObRD
t
.
. LASER
TIME
TRIGGER
BASE
-
0
6-
rn m
SI,
REFERENCE q-to-N
I -
-
1-
q-lo-N OUTPUT
~
TIME+
Figure 4. Waveforms for measurement system latches are transferred simultaneously to the 1/0 register of the DDP after the processor has executed an ENTER DATA command and an external timing signal is received. For these applications, this timing signal is the same as that used to clear the counters. Because of the large charge packet injected into the q-to-N during an intense light pulse, the output counts will normally be produced at the full-scale frequency and will continue after the laser pulse is over until the input charge has been nulled. The total number of counts must be delivered within one integration period of the time base for meaningful data to be obtained. This situation is easily attained by either changing the time base (which would change the laser repetition rate) or, preferably, by changing the current (or charge) sensitivity of the q-to-N converter or the gain of the photomultiplier tube. During the integration period, the counters integrate both signal and background. To derive the signal values alone, background values, which are obtained and stored in memory during a background cycle when the source radiation is blocked, are automatically subtracted from the signal plus background values by the processor. The background corrected values are then ratioed.
RESULTS AND DISCUSSION To establish that the simultaneous, dual beam, ratio system has the ability to correct for large variations in light intensity, a tungsten halogen lamp was operated directly off line voltage. The detectors were unfrosted 1P39 phototubes and the glass beamsplitter was used. Measurements were made at 500 nm. The results are shown in Figure 5. The data were obtained by taking 50 consecutive points, each representing the ratio in transmittance of the average relative intensities in the reference and sample channels over a 1-ms integration period. Then, another set of 50 data points was taken and added to the first set, Le., the 51st point was added to the lst, the 52d to the 2d, etc. This process was repeated 100 times. Each data value in Figure 5 is therefore the average of 100 measurements of 1ms each. By making the measurements in this manner, the 120-Hz ripple is clearly seen. All 50 points are not shown in Figure 5 . The single beam data (top curve) were obtained by holding the reference channel signal constant which was easily accomplished by increasing the charge sensitivity of the reference q-to-N until the output was continually at the overrange frequency. As can be seen, the ripple on the lamp is 120 Hz as expected for a 6 0 - H ac ~ supply and the magnitude of the ripple is about 8%of the total signal. By switching to double beam operation, i.e., bringing the reference q-to-N on scale so that it is responsive to the changes in the lamp radiance, the results shown in the lower curve of Figure 5 were obtained. Although the source intensity is changing by 8%,the ratio output of the measurement system
U
E 101
i
8 100 99
,
*
,
~
~
,
.
,
.
"
,
*
,
X
X
,
I
,
"
:
~
X
:
*
:
X
:
"
~
:
"
I
~
~
,
~
~
~
:
TIME ( 2 m S I D I V )
Figure 5. Comparison of single-beam and double-beam modes of operation for tungsten halogen lamp shows variations of only 0.1%. Consequently, a high intensity source with large ripple can become a viable source for even fast kinetic determinations. Although many double beam systems would yield an improvement in this case, the simultaneous acquisition of reference and sample information allows the source to be operated directly off the line without any signficant problems due to output ripple. To demonstrate that the simultaneous dual beam ratiometric measurement system is capable of correcting for wide variation in source intensity from a pulsed light source, the performance of the system with a programmed deuterium hollow cathode lamp was evaluated. Measurements were taken at 240 nm with a quartz plate beamsplitter and P M tubes using a programming sequence for the Iamp of 10 ms ON at a given current, then 90 ms OFF, then 10 ms ON, etc., for 10 ON/OFF periods. Each data point then was the sum of 10 pulses of 10-ms duration each. The output intensity of the Dz hollow cathode lamp increased linearly with lamp peak current from 60 mA to at least 410 mA. However, for over a 3-fold change in source intensity, the ratio varied by only 0.2% RSD which demonstrates that the system is very insensitive to changes in source radiance which occur before the beamsplitter. It should be noted that a t currents lower than 100 mA, the ratio tended to decrease as the current decreased. This was probably caused by shifts of the output radiation on the P M tubes because of changes in the physical shape or location of the lamp discharge. Frosted P M tubes were not used when these data were obtained. To illustrate the correction attainable for pulse-to-pulse variations in the output of short-duration pulsed lasers, data were recorded on two laser systems. Results obtained for the small Na laser are shown in Table I. The average values and % RSD's are for 10 determinations of 10 measurements each. The background was updated each time the repetition rate
ANALYTICAL CHEMISTRY, VOL. 48, NO. 11, SEPTEMBER 1976
1533
Table I. Results for Small Nitrogen Laser Repetition Reference channel Sample channel rate, Hz Countsa R S D , % Countsa RSD, o/o
1 10 100
4288 5858 258
10.81 2.96 7.53
2250 3139 134
11.08 3.06 7.11
a Average of 1 0 determinations of
Ratio (S/R) Ratioa
IISD, %
0.5216 0.5355 0.5470
0.65 0.26 1.37
REFERENCE CHANNEL ( 5 . 5 % R S O ) (Np LASER1
>
t
1 0 pulses each.
z c
RATIO ( 05 2 % R S D )
J 0.35
3 W
was changed because of the change in integration times. Otherwise, all other parameters were kept the same. Exposed film was placed in the light paths to decrease the incident intensity on the phototubes to prevent saturation effects. For 1-Hz and 10-Hz repetition rates, the ratio precision shows more than an order-of-magnitude improvement over that of the single-channel measurements. For 100 Hz, the drop to a fivefold improvement likely results because the background has become significant with respect to the signal. The decreased signal at 100 Hz can be attributed to the depletion of the number of N2 molecules available for excitation during subsequent breakdowns because of the buildup of nitrogen in metastable states (23)from previous excitations. At high repetition rates, these excited molecules are not efficiently swept out of the plasma tuhe before the next firing. Conversely, at low repetition rates, e.g., 1Hz, many of the ionized species are swept out between pulses causing an effective increase in tube resistance. Therefore, lasing is less efficient with a resultant decrease in signal and precision relative to 10 Hz for which the pressure was nominally adjusted. In most applications, the nitrogen laser is operated at 10 Hz. As the repetition rate is varied, the shape as well as the intensity of the laser output can visually be seen to change by observing the radiation with an agent which will fluoresce under 337-nm irradiation. Consequently, the 5%change in the ratio between measurements taken a t 1Hz and 100 Hz is not unexpected because different portions of the photocathodes and beamsplitter are illuminated. However, in normal applications, the repetition rate would either be constant or not be changed over a wide range. The performance of the ratiometric system with the nitrogen pumped dye laser was evaluated using rhodamine 6G and several cavity configurations. No monochromator was used. The output of the dye laser was attenuated by a 3.0 OD neutral density filter before the beamsplitter module. An additional 1.0 OD filter was placed between the quartz plate beamsplitter and the phototube detector for the light passing through the beamsplitter. The data are presented in Table 11. Reference beam, sample beam, and ratio information is given for broadband lasing and for tuned lasing. Data are presented for the average of 30 determinations with each determination being either the average of 1pulse or of 10 pulses.
? 4:
L W
SAMPLE CHANNEL 15.3 % R S D ) (ANTHRACENE FLUORESCENCE)
I
I
I
IO
20
30
TIME (SECI
Figure 6. Fluorescencemeasurements on anthracene using a nitrogen
laser excitation source
Although the precision for the single beam data improves when the number of pulses averaged is increased by 10 times, the improvement is not a factor of di6as would be expected if the variations were totally random. This is because the output of the laser exhibits drift. I-lowever,the ratio precision does show a di6 improvement, indicating that the ratio compensates for the drift in the laser output as well as the pulse-to-pulse fluctuations. The ratio for broadband lasing with an aluminized reflector shows a factor of 6 improvement over single channel precision for 1pulse per determination and a factor of 12 for 10 pulses per determination. Using the grating perpendicular to the cavity optical axis (zero order) to obtain broadband lasing, the improvements are 13 and 17, respectively. The increase in precision may be attributed to the vertical polarization of the laser beam a t zero order caused by the grating rulings. For the tuned cavity, the improvement factors are not as good as for broadband lasing. This may be due to the fact that about 50% of the total intensity is reflected off the grating a t zero and is vertically polarized. Compensation for fluctuations of the laser output is obviously dependent upon the properties of the beamsplitter and the polarization of the radiation. In this application, the system performs best for nonpolarized or vertically polarized light. The small nitrogen laser, with its ultraviolet emission at 337.1 nm, is an effective excitation source for a broad range of fluorescent compounds and, in particular, for the enzyme cofactor NADH (7) which has an absorption maximum a t 340 nm with a fluorescence emission maximum at 460 nm. Because of the desire to use the N2 laser for molecular fluorescence
Table 11. Results for Rhodamine 6G Dye Lasera Reference channel Cavity characteristics
Broadband Reflecting mirrorb Reflecting mirrorc Grating (zero order)b Grating (zero order)c Tuned (590 nm) Gratingb Gratin g C 10-Hz laser repetition rate. each. Q
1534
Sample channel
Ratio (S/R)
Counts
RSD, %
Counts
RSD, %
Ratio
RSD, %
4176 3684 4425 4301
3.24 2.43 2.64 1.18
2289 1982 2666 2588
3.34 2.41 2.58 1.17
0.5481 0.5380 0.6024 0.6017
0.56 0.20 0.20 0.07
4503 2.30 2540 2.37 0.5642 1.04 4408 1.31 2496 1.21 0.5662 0.37 b Average of 30 determinations of 1 pulse each. CAverage of 30 determinations of 1 0 pulses
ANALYTICAL CHEMISTRY, VOL. 48, NO. 11, SEPTEMBER 1976
Table 111. Comparison of Long-Term Stability Average d r i f t (% T/h)a
Optical feedbackb Ratiometric modec
Ref channel
Sample channel
Ratio readout
0.005 0.15
0.15 0.15
...
0.008 U M e a s u r e m e n t s t a k e n o v e r a 14-h p e r i o d . b100-ms R l C l t i m e c o n s t a n t . = C o n s t a n t v o l t a g e on t u n g s t e n l a m p .
determinations, the ability of the simultaneous ratiometric system to correct for pulse-to-pulse variations during fluorescence measurements was important. T o evaluate the performance, 30" front-face fluorescence measurements were made on a solution of anthracene in cyclohexane at an emission maximum of about 400 nm. A quartz plate beamsplitter was used to direct a portion of the laser output to the 1P39 reference channel phototube with a 2.5 OD neutral density filter before it to prevent detector saturation. A photomultiplier was used to detect the fluorescence emission signal isolated by the monochromator. Measurements were taken over a period of 30 s with a laser repetition rate of 10 Hz. The results are shown in Figure 6. Each data point is the average of 10 measurements and therefore represents a time span of 1 s. The fluorescencesignal is seen to follow the variations in the laser intensity as expected if the excited electronic state is not near saturation (8).The precision of the ratio for the 30 points (300 pulses) shows an order-of-magnitude improvement over the reference or sample channel precision, demonstrating that the ratiometric system works very well in this application. Comparison of Optical Feedback and Ratio Modes. With a nominal 6.0 V on the CM-1630 tungsten lamp, the performance of the optical feedback system shown in Figure 2 was compared to that of the split beam ratio measurement system. Measurements for both modes of operation were made at 550 nm with a 1-nm bandpass. A quartz plate beamsplitter and frosted P M tubes were used. A summary of the results of a long-term drift comparison is given in Table 111. The drifts were measured from data plots on the strip chart recorder and were unidirectional, in general, after about a 4-h warm-up period. No special temperature control procedures were used. In the optical feedback mode, the readout of interest is the sample channel value and, in the ratiometric mode, the ratio readout. The ratiometric mode shows approximately a factor of 20 improvement in long-term stability over that of optical feedback. As expected, the reference channel stability for optical feedback is greater than for the ratiometric mode since this channel is the one providing the feedback signal to the lamp. However, there is no improvement in the sample channel value compared to that seen for constant voltage on the tungsten lamp. The reason for this is not obvious. A possible explanation is that as the voltage on the lamp is changed by the feedback to compensate for effects such as filament aging, the source image may be altered, thereby causing increased differential response of the tubes, particularly since any shift will cause the source image to move in opposite directions with respect to the two photocathodes because of the system configuration. The longterm stability of the ratiometric system compares favorably with that reported by Kirkland (24) for a split-beam system using phototube detectors and logarithmic amplifiers and to that reported by Pardue and Rodriguez ( 2 1 ) for an optical feedback system. Optical feedback systems have shown good results for low wattage sources with low frequency noise characteristics. However, to correct for higher frequency noise, such as 60 Hz, the feedback amplifier time constant must be decreased to
provide gain for these higher frequency components. For a source such as a tungsten lamp, the phase shift of the feedback loop may approach 360' as the feedback time constant is decreased because of the filament response time. The net result may be an increase in the noise level on the lamp. The short-term noise effects of changing the RC time constant of the feedback amplifier were investigated for the optical feedback mode. Every 2 ms, a data point representing a 1-ms integration was taken until 25 points were accumulated. One hundred ensembles of 25 points each were then averaged. Therefore, each value was an average of 100 determinations. For time constants of 100 ms and 10 ms, the sample channel signal was essentially constant. However, as the time constant was further reduced to 1ms, the output begins to show a 60-Hz ripple of about 0.5%of the full scale signal. Further reduction to 100 ps resulted in a 4%variation. So, as the amplifier time constant is decreased, increased amplification is provided for noise present in the system and it is not surprising that 60 Hz should predominate. An 8%of full scale (Ih V peak-to-peak), 60-Hz noise signal was then purposely added to the tungsten lamp. Now, with the 100-ms time constant, over a 1%peak-to-peak variation in intensity was seen. Some compensation for the impressed noise was obtained since the sample channel variation without OF is about 3%. Reducing the time constant in an attempt to correct for this fluctuation did not succeed and, in fact, the noise increased in amplitude for 1 ms and 100 ps time constants. The simple optical feedback circuit shown in Figure 2 is not capable of effectively correcting for noise signals which are faster than the frequency response of the lamp. In fact, as the RC time constant of the feedback amplifier is reduced, the noise signals are actually amplified by the increased gain of the feedback network. Therefore, the noise on the source can be enhanced over that which was initially present. The conditions were kept the same (noise added) except that the simultaneous ratiometric mode was selected. The reference and sample channels showed about 3% variations at 60 Hz. However, the ratio shows a RSD of about 0.1%, a limit predicted on the basis of photon statistics. Therefore, it is seen that the simultaneous dual beam, ratio mode has effectively corrected for the source noise, eliminating source flicker as a photometric error source. The optical feedback system is not capable of eliminating this flicker, at least not without making some fairly complex modifications to the feedback loop.
ACKNOWLEDGMENT The authors express their gratitude to Mark Simmons for his valuable assistance with the laser systems. LITERATURE CITED (1) (2) (3) (4)
J. D: Winefordner and T. J. Vickers, Anal. Chem., 46, 192 R (1974). J. R. Allkins, Anal. Chem., 47, 752 A (1975). K. R . O'Keefe and H. V. Malmstadt, Anal. Chem., 47, 707 (1975). T E. Hewitt and H. L. Pardue, Clin. Chem. ( Winston-Salem, N.C.), 21, 249 (1975). (5) L. M. Fraser and J. D. Winefordner, Anal. Chem., 44, 1444 (1972). (6) M. P. Bratzel, R. M. Dagnall, and J. D. Winefordner, Anal. Chim. Acta, 5 2 , 157 (1970). (7) M. J. Simmons, J. D. Defreese, and H. V. Malmstadt, Paper No. 157, Federation of Analytical Chemistry and Spectroscopy Societies Second National Meeting, Indianapolis, Ind., October 1975. (8) 6.Smith, F. Plankey, N. Ornenetto, L. Hart, and J. D. Winefordner, Spectfochim. Acta, Part A, 30, 1459 (1974). (9) "RCA Photornultlplier Manual", PT-61, RCA Corp., Harrison, N.J., 1970. (IO) H. L. Pardue and S. N. Derning, Anal. Chem., 41, 986 (1969). (11) H. L. Pardue and P. A. Rodriguez, Anal. Chem., 39,901 (1967). (12) P. A. Loach and R. J. Loyd, Anal. Chem., 38, 1709 (1966). (13) L. A. Rosenthal, Rev. Sci. lnstrum., 36, 1329 (1965). (14) GCAIMcPherson UV-VISIBLE Light Source, Model EU-701-50.
ANALYTICAL CHEMISTRY, VOL. 48, NO. 11, SEPTEMBER 1976
1535
(15) 6. Chance, D. Mayer, N. Graham, and V. Legallais, Rev. Sci. lnstrum., 41, 111 (1970). (16) J. D. Defreese, T. A. Woodruff, and H. V. Malmstadt, Anal. Chem., 46, 1471 (1974). (17) D. Harrington and H. V. Malrnstadt, Am. Lab., 6 (3), 33 (1974). (18) G. Sauerbrey, Appl. Opt., 11, 2576 (1972). (19) T. A. Woodruff and H. V . Malmstadt, Anal. Chem., 46, 1162 (1974). (20) R. P. Gregory IV, J. A w r y , 6.W. Renoe, P. C. Dryden, and H. V. Malmstadt, Clln. Chem. ( Winston-Salem, N.C.), 20, 950 (1974). (21) P. C. Dryden and H. V. Malmstadt, Paper NO. 29, Pittsburgh Conference
on Analytical Chemistry and Applied Spectroscopy,Cleveland, Ohio, March
1975. (22) T. A. Woodruff
and H. V . Malmstadt, Anal. Chem., 46, 2141 (1974).
(23) L. E. S. Mathlas and J. T. Parker, Appl. Phys. Lett., 3, 16 (1963). (24) J. J. Kirkland. Anal. Chem., 40, 391 (1968).
RECEIVEDfor review November 17,1975. Accepted June 14, 1976. Presented in part at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 1975. This work was supported in part by NSF Grant GP18910, and also in part by an American Chemical Society, Division of Analytical Chemistry Fellowship (to JDD) sponsored by the Procter and Gamble Company.
Spectrophotometric Determination of Hydrazides with 2,3Dichloro- 1,4-naphthoquinone J. A. Plairier,* J. G. Van Damme, and R. E. De Neve Laqoratorium voor Galenische en Magistrale Farmacie, Vrue Universiteit Brussel, Farrnaceutisch lnstituut, Paardenstraat, 67 B- 1640 St. Genesius Rode, Belgium
A new colorimetric method for hydrazides, using alkaline 2,3-dlchloro-l,4-naphthoqulnone is tested. The method is specific for hydrazides not substituted on the 0 N atom, excluding hydrazine, hydrazone derivatives, ureum and semicarbazide. Aromatic hydrazides conform to Beer's law from 3.10-$ to 2.10-$ molar. Dihydrazide derivatives can only qualitatively be determined except oxalic acid dihydrazide. As the reaction mechanism, by inspecting the uv spectra, is proposed the condensation between a ketone and a hydrazide to a hydrazone, tautomerizing in alkaline medium.
The recent widespread interest in carboxylic acid hydrazides for a variety of applications, and the subsequent growing commercial importance of several acid hydrazides ( I ) led to the search for an analytical method, universally applicable, for the determination of these hydrazides. Various methods have been proposed, especially for isonicotinic acid hydrazide (1"). Most of the proposed reactions are based on the reducing properties of the hydrazide function: one of the earliest methods was a titrimetric one with nitrous acid (2).Iodometric ( 3 , 4 )and bromometric ( 5 , 6 )titrations are reported, and also nonaqueous titrimetric procedures using perchloric acid in acetic acid (2,4,7,8). These methods, however, are not specific for the hydrazide function and cannot make distinction between a hydrazide and its hydrazone derivative (9). A volumetric determination of nitrogen after treatment with iodate (10) has been used for a variety of hydrazides. Several years ago, a fluorimetric method using Rubin's (11) cyanogen bromide reaction, had been reported. Although sensitive, the method is not specific as it is based on the reactivity of the pyridine ring. Colorimetric methods have been described, involving reaction between the hydrazides and an aldehyde- or keto group: a) p -dimethylaminobenzaldehyde forms yellow hydrazones (12) with hydrazine and nonsubstituted hydrazides. b) Alkaline 1,2-naphthoquinone-4-sulfonate (13, 1 4 ) forms orange-red reaction products with hydrazides, but also with methylene and amino groups (15, 16). In conclusion, it can be stated that none of the above mentioned reactions is specific for the hydrazide group. The method presented here is based on an observation of Van Damme and De NBve ( I 7) that three out of five tested 1536
*
naphthoquinones give a blue color in alkaline solution with isonicotinic acid hydrazide. As napthoquinones are unstable in alkaline medium, they extracted the colored reaction products with amyl alcohol. This study was undertaken to determine applicability and limitations of that method, for quantitative determination of acid hydrazides by means of a naphthoquinone. EXPERIMENTAL Products. All products used are commercially available: salicylic acid hydrazide (Aldrich); nicotinic acid hydrazide (Aldrich); isonicotinic acid hydrazide (Aldrich); phenylacetic acid hydrazide (Aldrich); benzoic acid hydrazide (Schuchardt);butyric acid hydrazide (Aldrich); oxamic acid hydrazide (Fluka); acetic acid hydrazide (Aldrich); oxalic acid dihydrazide (Aldrich); glutaric acid dihydrazide (Aldrich); succinic acid dihydrazide (Aldrich). The following products are also tested: hydrazine.2 HC1 (Merck); phenylhydrazineeHC1 (U.C.B.); hydroxylamine.HC1 (Carlo Erba); acetamide (U.C.B.); ureum (Carbo Erba); semicarbazidemHC1 (Analar). 2,3-Dichloro-1,4-napthoquinone (phygon) (Fluka): a saturated solution in petroleum ether was used. The choice of the solvent will be discussed later. Ammonia-ammonium chloride buffers were prepared a t sufficiently high buffer capacity in the pH range 8-12 by neutralizing NH3 (0.1-1 M) with HC1. Apparatus. All spectra were recorded with a Perkin-Elmer (124) double beam spectrophotometer, and all absorbance measurements, with a Vitratron U.C. 200 S. For pH measurements, a pH meter (Radiometer Copenhagen) was used. The aliquots were shaken in a Gerhardt horizontal shaker a t 340 movements/min. Methodology. All the hydrazides were dissolved in water and diluted to adequate concentration. To 5 ml of these solutions were added 4 ml of a saturated phygon in petroleum ether solution and 1 ml of buffer solution. The mixture was shaken. Conditions of pH, ionic strength, stability, and time, giving maximum color development were investigated. After separation of the layers, the absorption spectra were recorded. Quantitative determinations were made in 1-cm cuvettes at the appropriate wavelengths using the reagents (without hydrazides) as blanks. Conformance to Beer's law and sensitivity of the method were investigated.
RESULTS AND DISCUSSION Qualitative Determination. Phygon. Naphthoquinones are unstable in alkaline medium. Phygon was chosen as it is less sensitive than the other naphthoquinones. Soluent. If we used amyl alcohol, as investigated by Van Damme, the excess of the reagent, here phygon, a yellow re-
ANALYTICAL CHEMISTRY, VOL.'48, NO. 11, SEPTEMBER 1 9 7 6