Table I V . Recovery Data for Chloride with Xylenol Orange
Table Ill. Beer’s Law for Chloride with Xylenol Orange CI -, ppm
A , 588 n m a
CI- present, pprn
8
2.0 0.45 ... 4.9 0.74 5300 7.4 0.95 4500 9.8 1.35 4800 12.3 1.72 4900 a A = 0.30to 0.40for the blank for the lowest concentrations. Values
2.9 5.8 7.9 9.8 11.6
CI- found, pprn 2.7 5.3
8.0 10.2 12.3
Error, % -6.9 -8.6 +1.3 4-4.1 +6.0
are somewhat uncertain because of a blank correction problem.
lenhhy, involving two extractions and a concentration step. Molar absorptivities are reported to be 21,300 a t 257 n m and 6,500 at 297 nm. These values are based apparently on the concentration of the mercury complex in the final chloroform solution. It is uncertain as to how the effective molar absorptivity for chloride would compare with the use of PPC as a complexing agent.
Reaction of Mercuric Chloride with Xylenol Orange. Measurement of chloride using this reagent is somewhat more involved t h a n with PPC. The molar absorptivity of the HgC12-X0 complex was found to be about 24,000 at 588 nm, somewhat lower than that reported for the complex with mercuric nitrate (8). The stoichiometry is reported to be 1:l which was also found to be the case in this work using HgClz. T h e color develops rapidly and is reasonably stable. The p H is critical and must be very close to 7.0. It is also necessary to have a high concentration of hexamethylenetetramine for maximum color. Chloride ion can be measured in the range of 2-14 ppm with Xylenol Orange. Beer’s law plots from data obtained in this procedure were not quite linear over the entire range measured. A straight line was obtained from 2-7
ppm C1- and a second straight line with somewhat greater slope resulted from 7-15 ppm. Beer’s law data are shown in Table I11 and recovery data presented in Table 1V. Precision is acceptable if reasonable care is used in obtaining the Beer’s law plot. The effective molar absorptivity for chloride at the higher concentrations is approximately 4800. A 1:l ethanol-water solvent is necessary in order for the blank to be acceptably low. Since Xylenol Orange forms colored complexes with a large number of metal ions (12), it would also be necessary to remove most metal ions prior to the development of color. Possible interferences of other anions were not evaluated for Xylenol Orange but would be expected to be about the same as with the use of phenolphthalein complexone. Received for review January 8, 1973. Accepted March 19, 1973. The authors express their appreciation to the Robert A. Welch Foundation of Houston, Texas for partial support of this research. This work was also supported in part by the Faculty Research Fund of Sam Houston State University. (12) D. F. Bok and M . G. Mellon,Anal. Chem., 42 ( 5 ) , 152R (1970)
Analog Cross Correlation Readout System for a Spectrometer Using a Silicon Protodiode Detector Gary Horlick and Edward G. Codding Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2 Canada
Smoothing, resolution enhancement, and differentiation of spectra can be carried out using cross correlation techniques (1-4). The application of a self scanning linear silicon photodiode array detector to a spectrometer (5) has facilitated the development of a real time analog cross correlation readout system. By utilizing the system presented here, all of the above operations can be performed on spectra right a t the spectrometer by cross correlation with an electronic waveform that replaces the conventional mechanical exit slit.
(1) A. Savitzkyand M . J . E. Golay, Anal. Chem., 36, 1627 (1964). (2) Jean Steinier, Yves Termonia, and Jules Deltown, Anal. Chem., 44, 1906 (1972). (3) Gary Horlick,Anal. Chem., 44,943 (1972). (4)W. Snelleman, T. C. Rains, K. W. Yee, H. D. Cook, and 0. Menis, Anal. Chem., 42,394 (1970). (5) Gary Horlick and Edward G. Codding, Anal. Chem., 45,1490 (1973).
EXPERIMENTAL Apparatus A block diagram of the cross correlation system is
shown in Figure L4.The monochromator, photodiode array detector, clock, and measurement system were as described in reference ( 5 ) . A slit width of 100 fim was used for all measurements. A spectral region of about 110 8, was detected by the array which is 0.256 inch long and is made up of 256 photodiodes. The array was scanned at a clock rate of 30.0 KHz so that a complete spectrum (110 8, region) was read out in 8.5 msec with a repetition time (time between start pulses) of 60 msec. The electronic gate waveforms which were cross correlated with the spectra (Figure 1B) were synthesized by the gate function generator which consisted of three delay monostable-gate monostable combinations and a system of operational amplifiers. In order to synchronize the gate waveforms with the spectral signal, the delay monostables were triggered by the photodiode array start pulse. Depending on the application, widths of the gate waveforms varied from 100 to 400 fisec which is equivalent to approximately 1.3 to 5 8, on the spectral scale. Procedure. To perform the cross correlation, the spectral signal from the photodiode array and a gate waveform were multiplied together (-17 multiplications per sec) using an analog multiplier
ANALYTICAL C H E M I S T R Y , VOL. 45, NO. 9, AUGUST 1973
1749
DIODE ARRAY DETECTOR
MONOCHROMATOR
ARRAY CLOCK AND MEASUREMENT
A
*
STRIP CHART
SCAN CONTROL
t
MULTIPLIER /INTEGRATOR
A
J -lov
H
100p,cc
JL
01
B Figure 1. A . block diagram of the cross correlation readout system: B, typical electronic waveforms for smoothing ( l ) , resolution enhancement (2),and differentiation (3)
A
1
3 8 2 b 3832 3 8 3 8
B
,
3844
i
-
bo74 bOPb b143 bIb3 A
C
Figure 2. A , smoothed spectrum of a Cr triplet; B, normal and resolution enhanced spectrum of a Mg triplet; and C, normal and first derivative spectrum of Ne in the 61 00 A region
(PAR model 230) as the monochromator was slowly scanned (0.2 A/sec). The products were integrated using the low pass filtered output of the multiplier with a time constant of 3 seconds. This is analogous to scanning a dispersed spectrum past a mechanical exit slit. In this case, the gate waveform amounts to an “electronic slit.” The ability to easily manipulate the shape of this electronic exit slit provides the unique versatility of the cross correlation readout system.
RESULTS AND DISCUSSION Smoothing. Spectra can be smoothed by cross correlation with the square pulse waveform shown in Figure 1B. The degree of smoothing is a function of the width of the pulse with respect to the spectral features being observed. This is illustrated in Figure 2A. The spectrum of the Cr triplet (4339.4, 4344.5, and 4351.8 A) has been cross correlated with square pulses of widths 200 psec (-2.5 A), 300 psec (-3.8 A), and 400 psec (-5.0 A). Note the loss of resolution with increased smoothing which is essentially analogous to the well known effects of increasing the conventional slit width. The smoothing waveform need not be restricted to square pulse waveforms. With modern triggerable waveform generators, the gate pulse could be triangular, 180” of a cosine wave, a sawtooth, or essentially any desired form.
1750
Resolution Enhancement. By changing the electronic gate to the second waveform shown in Figure 1B. the cross correlation read out system operates in a resolution enhancement mode. Cross correlation of a spectrum with this waveform cannot be defined precisely as either taking the second derivative or deconvolving the spectrum, but it is analogous to both these operations. The resolution enhancement effected by cross correlation with this waveform is illustrated with a Mg triplet (3829.3, 3832.3, and 3838.3 A) in Figure 2H. The upper spectrum was obtained using the square pulse waveform [Figure 1B ( l ) ]of width 100 psec (-1.3 A ) . The lower spectrum was obtained using the resolution enhancement waveform [Figure 1B (2)] in which the width of the central pulse and negative sidelobes was 100 psec and the amplitude of the sidelobes was one half that of the central pulse. The shoulder on the first peak is now clearly evident. The separation of the first two peaks as measured from the strip chart recording is 3.0 A and of the second two peaks is 6.0 A, both of which are in agreement with the reported values. The precision of these values as limited by reading error is about h0.2 A. The degree of resolution enhancement may be conveniently controlled by varying the relative amplitude of the sidelobes with respect to the central peak. It should also be noted that the width of the sidelobes and central peak is quite critical. If they are too wide, the spectrum is simply smoothed and negative sidelohes are added to the peaks. Differentiation. The first derivative of a spectrum can be obtained by cross correlation with the third waveform shown in Figure 1B. This is a rather square approximation to a first derivative response function and as such an exact first derivative is not obtained. Again it is a simple matter to control the degree of smoothing by controlling the width of the pulses. Some smoothing is always necessary in conjunction with differentiation to avoid a noisy result. A spectrum of five Ne lines measured in the normal mode of operation with a pulse width of 175 psec (-2 A) is shown in Figure 2C. The differentiated spectrum shown directly below it was obtained by cross correlation with the differentiating waveform. Both the positive and negative sidelobes were 175 psec wide and had equal amplitudes. Thus, this analog cross correlation system provides a unique and versatile approach to spectrometric readout. In addition, more complex cross correlations can be used to detect spectral patterns (6). These could be carried out with this system using a gate function that takes the form of a specific spectral pattern. Such a system would also be somewhat analogous to mask correlation spectrometers (7, 8) that use a characteristic mask in the exit focal plane to “gate out” specific spectral patterns. In a system developed with the approach presented in this paper. the mask could be replaced by a characteristic electronic gating waveform. Received for review October 24, 1972. Accepted January 29, 1973. Financial support by the University of Alberta and the National Research Council of Canada is gratefully acknowledged. (6) Gary Horlick, Anal. Chem., 45, 319 (1973). (7) J. H . Davies,Anab Chem.. 42(6). 1OlA (19701. (8) D. T. Williams and B. L. Kolity, Appl. Opt.. 7. 607 (1968)
ANALYTICAL C H E M I S T R Y , VOL. 45, NO. 9 , A U G U S T 1973
