the measurement of 17-hydroxycorticosteroids in urine specimens using HBS procedure. Finally, a diabetic urine with 4.9 grams of glucose per 100 ml was extracted with ethyl acetate after an overnight hydrolysis with @glucuronidase and 17-hydroxycorticosteroids were determined with HBS. As before, presence of glucose did not impede these determinations. None of the tested compounds exhibited any detectable absorbance when heated at 100 “C for 60 min with 0.6: 1 phosphoric acid-water alone. The absorbances for THF and THS were 87.5 and 96.2%, respectively, of that found with cortisol. T H E showed a 12% greater absorption than that observed for cortisol. The mechanism of reaction and the nature of the chromophore formed with HBS reagent and the dihydroxyacetone side chain of steroids would appear to be similar to the PorterSilber reaction-namely, a n acid-catalyzed Mattox rearrangement (9) followed by formation of a 20-keto steroidal 21monohydrazone. Cortisol (Comp. F), which contains a A4-3ketone group in ring A in addition to the 17,21-dihydroxy-20 keto structure, shows a n initial absorption a t 338 nm when treated with HBS reagent for 10 minutes a t 80 “C. The absorption maximum then shifts toward 350 nm over the next 30 minutes and remains stable thereafter. This would indicate that the formation of a 3-hydrazone precedes the formation of a 21-hydrazone, with the final product being a 3,21-dihydrazone. In the absence of a A4-3-keto group, for example, with 3cr,ll~,l7a,21-tetrahydroxy-5~-pregnan-2O-one (THF), such (9) V. R. Mattox, J. Amer. Chem. SOC.,74,4340(1952).
a shift in the absorption maximum is not observed. Similar behavior was also noted with cortisone, 11-desoxycortisol, and their respective tetrahydro derivatives. It should be mentioned that formation of a 3,21-dihydrazone is also postulated to occur by reaction of cortisol with the Porter-Silber reagent (IO). The 350-nm absorption maximum for cortisol is in excellent agreement with the reported 350-nm maximum for pyruvaldehyde-1 -phenylhydrazone ( I I ) , the simplest analog of a 20-keto steroid 21-hydrazone. We are currently investigating in detail the course of HBS reaction with cortisol and the constitution of the final product obtained. We have used the HBS procedure outlined here routinely in the analysis of urinary 17-hydroxycorticosteroids in clinical specimens for 3 years and it does appear to show certain advantages-viz., greater sensitivity and shorter time required to perform the analysis. These studies are the subject of a subsequent report. ACKNOWLEDGMENT
We gratefully acknowledge the technical assistance of Hasmukh Desai throughout this work. RECEIVEDfor review May 31, 1972. Accepted August 30, 1972. (10) R. H. Silber and C. C . Porter, in “Methods of Biochemical Analysis,” D. Glick, Ed., Vol. IV, Interscience, New York, N.Y., 1957, p 139. (11) R. E. Bowman and C. S. Franklin, J . Chern. SOC., 1957, 1583.
AIDS FOR ANAL’IVTICAL CHEMISTS Instrumental Artifacts in Differential Pulse Polarography Joseph H. Christie, Janet Osteryoung, and Robert A. Osteryoung Departments of Chemistry and Civil Engineering, Colorado State University, Fort Collins, Colo. 80521
PULSEPOLAROGRAPHY (1) is a well-established electrochemical technique whose analytical applications have not been widely exploited. Although differential pulse polarography is ideally suited to analysis (2-4), lack of availability of a n economical, reliable instrument has hindered acceptance of the method for routine analysis. Recently Princeton Applied Research, Inc., introduced their Model 174 Polarographic Analyzer which seemed to answer the need for a n instrument for analytical applications. With any new instrument, it is necessary t o evaluate the instrument o n well-behaved systems before application to routine analysis. It is especially important t o understand the effect of the various system parameters o n the response of the instrument. In our preliminary experiments with the (1) G. C. Barker and A. W. Gardner, A . E . R . E . Hurwell, C/R 2297 (1958). (2) E. P. Parry and R. A. Osteryoung, ANAL.CHEM., 37,1634 (1965). (3) Ibid., 36, 1366 (1964). (4) M. Devalenola and P. NangniGt, Tulu/~tu,15, 759 (1968). 210
Model 174, we noted a variation of the peak current with scan rate in the differential pulse mode. I n their Preliminary Instruction Manual ( 5 ) for the Model 174, PAR indicates that a distortion of the peak takes place at faster scan rates because of the long time constant of the memory circuits used and recommends that scan rates of 2 mV/sec or less be used in the differential pulse mode. Our discovery of this instrumental artifact led us to carry out a complete evaluation of the effect of the instrumental parameters-drop time, r ; scan rate, v ; pulse amplitude, AE; and current range-on the response of the instrument for various concentrations of Cd(I1). The concentration range chosen was one over which the electrochemistry is expected to be well-behaved: 8.3 X 10W6t02.6X 10W3F. A detailed analysis of our results indicates that the P A R Model 174 is suitable for analytical applications if used with “Instruction Manual, Polarcgraphic Analyzer, Model 174,” Princeton Applied Research Inc., Princeton, N. J., p V-21.
(5)
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
Table I. Dependence of Peak Current and Peak Potential on Cd(I1) Concentration and Scan Rate at Constant Pulse Amplitude and Drop Time. 2 mV/sec 1 mV/sec 5 mV/sec Scan rate 0.5 mV/sec E,, mV cs. Ep, mV cs. E,, m V T E,, mV cs. SCE iP, PA SCE ip, P A SCE i ~ MA , ip, P A SCE [Cd(II)IF 0,089 - 653 0.059 - 661 0.104 -645 0.133 - 645 8.3 X -654 0.174 - 663 0.255 0.302 -648 0.320 - 644 2.5 x 10-5 1.18 - 654 0.81 -668 1.38 - 648 1.47 - 642 1 . 1 x 10-4 3.30 - 653 2.31 - 664 3.93 - 648 4.20 - 646 3.0 x 10-4 5.29 - 664 7.76 - 656 8.99 - 650 - 642 6.8 X 9.66 32.3 -656 22.1 - 666 37.6 - 650 -644 2.6 X 40.2 Pulse amplitude: 5 mV; drop time: 2 sec. Table 11. Dependence of Peak Current and Peak Potential on Cd(I1) Concentration and Drop Time at Constant Pulse Amplitude and Scan Rate. 0.5 sec 1 sec 2 sec 5 sec Drop time E p , mV cs. E p , mV us. E,, mV L‘S. E,, mV cs. SCE ip, P A SCE [Cd(Wl F ip, P A SCE i, r A SCE ip, PA 0.145 653 645 0.047 641 0.069 644 0.104 8 . 3 x 10F - 648 0.435 - 656 0.302 0.203 - 643 2.5 x 10-5 0.135 - 643 1.99 -654 1.38 - 648 0.613 - 641 0.934 - 643 1 . 1 x 10-4 5.63 - 655 - 648 2.63 - 643 3.93 1.74 - 641 3.0 x 10-4 Pulse amplitude: 5 mV; scan rate: 1 mV/sec. an understanding of the effect of system parameters on system output. 4-
EXPERIMENTAL
Differential pulse polarograms were run on the Princeton Applied Research Model 174 Polarographic Analyzer used in conjunction with a Model 172 Drop Timer and Stand. The cell used was a Sargent H-cell. The DME was in the sample compartment of the H-cell; in the other compartment were a Pt-wire coil counter electrode and a Coleman saturated calomel reference electrode. The pulse polarograms were recorded on a Hewlett-Packard Model 7035B x-Yrecorder on a sensitivity of 1.O V/in. ( Y )and 0.1 V/in. (X); this Y range corresponds to 7 0 z full scale output (10 V) of the Model 174, the X range corresponds to a potential scale of 50 mV/in. when the potential range of the polarograph was 0.75 V. The materials used were reagent grade. A 0.481F stock solution (Solution I) of Cd(I1) was prepared by dissolving Cd(NO&. 4 H 2 0 in water. A more dilute solution (Solution 11) was prepared by a 1 :25 dilution of Solution I. The nitrogen used to deoxygenate the solutions was purified by passage over hot copper; the nitrogen stream was passed through 1F HCI before being passed through the cell. While polarograms were being run, nitrogen was oassed over the surface of the solution. Except for small annulus around the D M E to allow the capillary to move to dislodge the drops and to permit the escape of nitrogen, both compartments of the H-cell were sealed with rubber stoppers. Pulse polarograms at “zero scan rate” were recorded manually, recording the electrometer output and the current output as measured on a Fairchild Model 7050 Digital Voltmeter. The potential was set using the initial potential control on the polarograph. The current output was measured after several drops at the same potential; this wait of several drops allowed the current memory to come up to full charge. After blanks were run on 1F HC1, Cd(I1) was added stepwise to the test solution (25 ml) by adding 10-100 p1 amounts of the stock solutions.
a
RESULTS
The data obtained are collected in significant groups in several tables and figures.
04-
03
30
2-
-
02 -
01-
f
SCAN RATE, mV/SEC
Figure 1. Dependence of peak current on scan rate for various Cd(I1) concentrations. A E = 5mV; T = 2sec 2.6 X b. 6.8 X C. 3.0 x d. 1.1 x e . 2.5 X .f. 8.3 X a.
10-3F 10-‘F 10-4~ 10-4~ 10+F 10-6F
Data were obtained for a 5-mV pulse amplitude and a 2-sec drop time for all concentrations and for scan rates of 0.5, 1, 2, and 5 mV/sec. These data are shown in Table I. This range of scan rates extends from one setting above the highest recommended by PAR to the lowest setting compatible with a reasonable analysis time. (At 0.5 mV/sec, 10 min are required to scan a 300 mV potential range.) The drop time of 2 sec was chosen as a n intermediate value. The smallest
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
*
211
Table 111. Dependence of Peak Current and Peak Potential on Scan Rate and Drop Time at Constant Cd(I1) Concentration and Pulse Amplitude. Drop time 0.5 sec ___ 1 sec 2 sec 5 sec E,, mV cs. Scan rate E,, mV cs. E,, mV cs. E,, mV cs. SCE i,, P A SCE mVjsec i, P A SCE ip, P A SCE in FA 2.54 -640 4.06 - 639 0 1.60 - 640 7.56 - 638 4.18 - 644 2.64 -644 7.71 - 644 ll7 1.70 -644 3.93 -648 5.63 - 655 2.63 -643 1 1.74 - 641 a Pulse amplitude: 5 mV; Cd(1I) concentration: 3.0 X lO-'F. Table IV. Dependence of Peak Current and Peak Potential on Pulse Amplitude and Scan Rate at Constant Cd(I1) Concentration and Drop Time. 2 mV/sec 0 mVisec 0.5 mV/sec 1 mVjsec Scan rate 5 mVjsec 1-u AE E,, mV os. E,, mV os. 1 u' E,, mV cs. E,, mV cs. En, mV cs. mV i, pA SCE i, FA SCE i, pA SCE ip,lA SCE i, pA SCE 2.31 -664 3.30 -653 3.93 -648 5 0.097 4.06 -639 4.20 -646 4.20 -660 6.30 -649 10 0,193 7.49 -643 8.09 -639 8.04 -637 14.6 -643 9.80 -654 25 0.453 17.5 -637 18.5 -632 18.8 -629 25.4 -633 17.3 -643 50 0.750 28.9 -626 30.9 -616 30.4 -620 28.6 -628 36.6 -616 100 0.960 39.0 -604 39.9 -596 39.7 -590 a Drop time: 2 sec; Cd (11) concentration: 3.0 X lO-'F.
+
g
t 4
I+; . . , ' +I
-0 -0
a
3
-0
'*
- 5
0
- -I --I
--I -I
1
-
I
01 0
I
L
1
2
3
4
2.0
IO
5
LOG C ( p M )
SCAN RATE, mV/SEC
Figure 3. Log i, cs. log [Cd(II)] for various scan rates. AE = 5 mV; 7 = 2 sec. (Note the shift in the log i, scales)
Figure 2. Dependence of peak current on scan rate. [Cd(II)] = 3.0 X lO-'F; AE = 5mV; 7 = 2sec
a.
pulse amplitude available (5 mV) was used to give a severe test of system performance, i.e., smallest response, and to give the sharpest, best-resolved peaks. The peak currents are plotted as a function of scan rate in Figure 1 for scan rates of 0.5, 1, and 2 mV/sec. Over this range, the peak current is a linear function of the scan rate. The point at 5 mVlsec lies above the extrapolation of the line determined by the first three points. A typical set of data for the entire range of scan rate is shown in Figure 2. Note that the point at zero scan rate lies below the straight line determined by the points at 0.5, 1 , and 2 mV/sec. In Figure 3 are plotted log calibration curves relating the peak current to the concentration at constant pulse amplitude 212
3.0
Y
=
5 mV/sec
h. v = 2 mV/sec c'.
v = 1 mVjsec
d. v
=
0.5
mV/sec
and drop time for scan rates of 0.5, 1, 2 , and 5 mVkec. The first and last points seem to be off the straight line in each case; this may be seen more clearly in the linear plots shown in Figure 4. The dependence of peak current on concentration and drop time at constant pulse amplitude ( 5 mV) and scan rate (1 mV/sec) is shown in Table 11. It should be noted that the peak potential is not independent of drop time. The peak current for each concentration is shown as a function of T ~
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
/
~
[Cd
0
a)] .mM
p
(DROP
TIME)”’,(SEC?’’
[Cd5)]mM
Figure 4. Linear calibration curves: i, us. [Cd(II)]. AE = 5 mV; 7 = 2 sec; v = 1 rnV/sec
Figure 6. Peak current as a function of r2j3 (proportional to electrode area). AE = 5 mV; [Cd(II)] = 3.0 X 10-4F a.
6.
v = o VT
=
1
e
Table V. Dependence of Peak Current and Peak Potential on Current Scale at Constant Drop Time, Scan Rate, Pulse Amplitude, and Cd(I1) Concentration. Current scale, pA i,, !.LA E,, mV cs. SCE
6
0.2 0.104 0.5 0.101 1 .0 0.103 uCd(II) concentration: 8.3 X F. Pulse mV. Scan rate: 1 mV/sec. Drop time: 2 sec.
4
3
.u“ 4
- 645 - 644 - 644
amplitude: 5
F o r a concentration of 0.30mF Cd(I1) and a constant 2-sec drop time, pulse polarograms were obtained for pulse amplitudes of 5 , 10, 25, 50, and 100 mV and scan rates of 0, 0.5, 1, 2, and 5 mV/sec. These data are shown in Table IV. According to Parry and Osteryoung ( 2 ) , the peak current should
2
0
(E?;).
Such be a linear function of --, 1 - u where u = exp - l+u a plot is shown in Figure 7 . Note that i, is a linear function 1-u of -only at slower sweep rates. The scan rate dependence l+u of the peak current is shown in Figure 8. Note the similarity to Figure 1 and note that the point a t zero scan rate lies below the straight line determined by the other three points. Parry and Osteryoung ( 2 ) also point out that the peak potential is given by
(DROP TIME~’:(SEC~”
Figure 5. Peak current as a function of 7 2 1 3 (proportional to electrode area) for various Cd(I1) concentrations. AE = 5 mV; v = 1 mV/sec 3.0 x 1 0 - 4 ~ h. 1.1 x 10-4F C. 2.5 x 1 0 - 5 ~ d. 8.3 X 10-6F
a.
in Figure 5. At constant scan rate, the peak current is not directly proportional to the electrode area as it should be if there were no distortion. At zero scan rate, the current is directly proportional to the electrode area as shown in Figure 6u; a similar result also obtains if a constant scan rate times drop time product is used, cf. Figure 6h. Under these circumstances the peak potential is also independent of scan rate, as shown in Table 111.
E,
= Ell?
- AEi2
(1)
Figure 9 shows plots of EDL‘S. A E for several scan rates. The plots are approximately linear, except for the points at A E = 5 mV and the slopes are all approximately 0.5 as predicted. The data in Table V show thdt the peak current and the peak potential are independent of the current scale o n which the pulse polarogram is recorded, all other instrumental parameters being kept constant.
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
213
40-
30
-700
-
-650
-
-
4
I.
,420-
W
u v)
-
k?
>
10 -
E. W"
-600
I
-
-0
1 - u Figure 7. Dependence of peak current on _ _ for various l + u
L
a.
b.
c. d. e.
AE,mV
mV/sec 0.5 mV/sec 1 mV/sec 2 mV/sec 5 mV/sec 0
Figure 9. Dependence of peak potential on pulse amplitude for various scan rates. [Cd (II)] = 3.0 X 1 O - T ; 7 = Zsec a. 0 mV/sec h. 0.5 mV/sec e. 1mV/sec d. 2 mV/sec e. 5 mV/sec
I
\
Table VI.
Calculated Values for Peak Current and Peak Potential for Cd(I1) Reduction' Scan rate, Relative peak AEp,mV Obsd mV/sec current Calcd 0 0.5 1 .o
O T
2
00-*
S X N RATE, mV/SEC
Figure 8. Dependence of peak current am scan rate for various pulse amplitudes. [lCd(II)] = 3.0 X 1 O - F ; 7 = Zsec a.
100 mV
b.
50mV
e. 25 mV d. 1 0 m V e. 5 m V
DISCUSSION
The deviations from theoretical behavior that have been observed can be explained by the slow response of the memory in the Model 174. The peaks are progressively shifted to more cathodic potentials and the peak current is diminished as the scan rate is increased (at constant drop time) or the 214
100
50
scan rates. T = 2 sec; [Cd(II)] = 3.0 X 10-4F. The origin for each curve has been shifted right by 0.05
1.oo 0.90 0.81
2.0 0.66 = 2sec; AE = 5mV.
0 -5.4
-5
-9.2 -13.9
-15
-9
drop time is increased (at constant scan rate). As shown in Table I11 and Figure 66, the peak potential remains constant and the peak current is directly proportional to the electrode area if the scan rate and the drop time are varied simultaneously, keeping their product constant. If this product is kept constant, the potential change per drop is the same and the distortion should be the same for each polarogram. Since the contents of the memory increase by only a fraction of the difference between the input and the previous contents on the memory (9,the more closely spaced the points and the more points that have been measured, the closer to the true value of the current will be the contents of the memory at the peak. Since the memory time constant is 110 msec, and the memory is connected for only 16.7 msec, the fractional increase in memory contents is y = exp(16.7/110) = 0.141. One can immediately write for the contents of the memory after the kth drop, Mk = Y ( h - M M )
+
Mk-1
(2)
where M j ( j = 1, . . . , k ) is the contents of the memory after the jth pulse and i f is the true current due to the j t h pulse. We have calculated the relative peak current and relative
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
Table VII. Analysis of Scan Rate Dependence of Peak Current for Variable Pulse Amplitude. 1 Ai, _ _
Ai,] Au AE, mV 5
10 25 50 100
a
PA-sec/mV 0.60 1.15
2.60 3.35
i, Av sec/mV 0.143 0.142 0.140 0.110
A E p i ~mV , 50 49 53 64
AE,~ zAip i, Au’ sec 7.2 7.0 7.4
Table VIII. Analysis of Scan Rate Dependence of Peak Current at Constant Pulse Amplitude.
Ai, _.
AV’
pA-sec/mV LCD(W1 F 0.0165 8.3 X 0.0405 2.5 x 10-5 0.185 1 . 1 x 10-4 0.605 3 . 0 x 10-4 1.25 6.8 X 5.30 2.6 x 10-3 a Pulse amplitude: 5 mV; drop time: 2 sec.
7.0 5.9
0.057 102 2.30 Cd(I1) concentration: 3.0 X 10-4 F; drop time: 2 sec.
peak potentials using this recursion relation; the results for Cd(I1) for v = 0, 0.5, 1.0, and 2.0 mV/sec, 7 = 2 sec, and AE = -5 mV are shown in Table VI. Note the excellent agreement with the calculated and observed shift in peak potentials. A similar argument can be used to explain the deviation from linearity shown in Figure 7 for large values of AE. F o r large AE the pulse polarographic peak is broader than a t smaller values. F o r a given scan rate and drop time, more points at which current flows have been passed through before reaching the peak for a large AE than for a smaller AE. The peak current for a large AE is therefore closer to the true value than for a smaller AE. The broader peaks a t large AE are also reflected in the smaller scan rate dependence of the peak current for large AE shown in Figure 8. In Table VI1 are shown the slopes Ai for the various values of AE used. Note that -2 passes Au AlJ through a maximum as the scan rate is increased. If these slopes are normalized by dividing by the peak current at u = 0.5 mV,‘sec, the data in Column 3 are obtained. (The peak current a t any other scan rate could be used, but the qualitative result would be the same.) The relative slopes for AE = 5 , 10, and 25 mV are essentially the same, with a decrease at 50 and 100 mV. I n Column 4 are shown the halfpeak widths, AE,,?, of the pulse polarograms a t u = 0.5 mV/ sec. The peaks are appreciably broader at 50 and 100 mV pulse amplitudes. Finally, in Column 5 are shown the products of the half-peak widths and the relative slopes. While these values are not constant, they are fairly close and their closeness supports our qualitdive argument that the distortion in the peak is related to the number of opportunities that the memory has had to sample a (non-zero)current. Similar calculations were performed on the scan rate dependence at constant pulse amplitude (Figure 1); the results are shown in Table VIII. The relative slopes in this case are constant, as they should be, since the peak widths are constant. The calculated slope, CL Table VI, is 0.17 sec/mV.
3
CONCLUSIONS
This work indicates that the peak current in differential pulse polarography using the Princeton Applied Research Model 174 is, in general, not independent of the scan rate used and that the peak current depends on the drop time and the pulse amplitude in a nontheoretical manner. Use of a computerized pulse polarographic system (6) which makes instantaneous current measurements, removes the
(6) H. E. Keller and R. A. Osteryoung, ANAL.CHEM.,43,342(1971).
1 Aip
i,ay’ sec/mV 0.15
0.13 0.13 0.14 0.13 0.13
instrumental artifacts reported in this work. Flato ( 7 ) indicates that with a Model 174 Pulse Polarograph, which had been specially modified so as t o have a 50-psec sampling interval, with a memory time constant of a few microseconds, similar results could be obtained--i.e., the instrumental artifacts disappeared; the peak amplitudes and peak position of a given sample were essentially constant with scan rate from a scan rate of 10 to 0.2 mV per second. The long time constants were included in this instrument by the manufacturer to improve the signal-to-noise ratio in very dilute solutions where the absence of the filtering and averaging provided would result in significantly noisier signals ( 7 ) . The use of a n integration-current measurement (sampling interval) method tends to even out random electronic noise which might be encountered in the instrument, but of itself does not contribute to the artifact problem. Rather, it is the small ratio of this sampling interval to the time constant of the memory circuit which results in these artifacts. Based on these results, several observations of importance for analysis can be made : (1) If the experimental parameters of scan rate, drop time, and pulse amplitude are kept constant, the peak current is directly proportional to the concentration of the electroactive species. (2) Analysis time can be shortened by using faster scan rates at the expense of a moderate loss in response, e.g., increase of scan rate from 0.5 mV/sec to 2 mVjsec decreases analysis time by a factor of 4, but decreases peak current by only about 20 %. (3) At faster scan rates, peak current will be increased by more than expected theoretically by increasing the pulse amplitude, e.g., greater than lox increase by increasing pulse amplitude from 5 to 100 mV. (4) For a given pulse amplitude, the peak current will be directly proportional to the drop area if a constant scan rate times drop time product is used. The existence of these instrumental artifacts does not preclude the use of this instrument in analysis, but rather reinforces the need for adherence to the principles of good analytical practice. The calibration and the analysis should be done under the same conditions, and the instrument should be understood before it is used. RECEIVED for review June 19, 1972. Accepted August 23, 1972. This work was supported, in part, by the National Science Foundation under Grant No. GP-31491, and by a Biological Sciences Support Grant from Colorado State University. (7) J. B. Flato, Princeton Applied Research Corporation, Princeton, N. J., privata communication, 1972.
ANALYTICAL CHEMISTRY, VOL. 45, NO. 1, JANUARY 1973
215
