1723
Anal. Chem. 1981, 53, 1723-1725
in Figure 2 is obtained. When a colored filter, C, is introduced between the scintillating light source and the detector, the solid line interacting a t PC is obtained. If discriminators d l and d2 are added to the system, the shaded areas rc and PO will represent total counts detected from the light source, with and without filter, respectively. A more detailed discussion is given by Ross (1). The concentration/diisplacementrelationship developed by
Ross
R = log (ro/rc) = kc
(1)
where R is the displacementfactor, ro is the count in the blank, rc is the count in the sample, k is the absorption constant for the system, and c is the molarity of the absorbing species, is used here with one significant change. The sensitivity factor, k', used in this study is empirical rather than derived from the theoretical principle used by Ross. This difference in the respective sensitivity factors is due to a combination of elements. In our instrumentation (a) the sample symmetrically surrounds the light source; (b) the light is detected by twin photomultipliers; and (c) the resulting signal is processed through coincidence-circuitry. It was not our intention to repeat the Ross experiment in its entirety; it was only necessary to demonstrate that we could generate adequate standiwd curves and compute k'values for a range of colored species. We found that a1 linear relationship between concentration ( c ) and displacement (R) exists for these solutions which ranged in color from orange to green. The k ' values for these salts are listed in Table I, which contains both experimental and calculated values. The expected decrease in sensitivity in going from the red to the blue end of the absorption spectrum is due to both the fluorescence spectrum of the scintillant and the response curve of the phototubes. Although there is no theoretical basis for comparing molar absorptivity (e) values with sensitivity factor (k? values, it is noteworthy that they are of the same order of magnitude, except in the case of chromium acetate. The apparent anomalous behavior of the latter may be due to its double absorption maxima shown in Figure 3. The advantages of thiri type of photometric measurement over conventional photometry were discussed in detail by Ross and include: (a) accuracy and precision limited only by statistical considerations of iradioactivity counting; (b) ability to handle very dilute to very concentrated samples with equal reliability; (c) stability of the light source; and (d) ability to feed digital output directly into a computer without analogue to digital conversion.
IW
sa0
\
/
I
0.2
110
410
1w
490
1110
e10
e10
Mo
e90
Wavebqih ( nm )
Flgure 3. Absorption spectra of K2Cr207,C ~ ( O A C ) ~and , Co(NO&.
The inherent limitations of our sample cell are comparable to those of the Ross system and are subject to the same corrective measures, e.g., the introduction of narrow band-pass filters around the light source to minimize interferences from other colored species. The loss in sensitivity resulting from the decreased light output is readily compensated for in either system by increasing the concentration of the radioisotope in the light source. Our light source/sample cell, however, has its own unique advantages: (a) it requires no modification of the LSC; (b) it is inexpensive and simple to fabricate; (c) its practical visible range can be altered simply by changing to a different scintillant system; and (d) it should have applications in numerous instances, such as in differential reaction rate studies, where it is very difficult to use conventional photometry. The most attractive feature of this light source/sample cell is that it permits high-precision measurements to be made a t low cost with an unaltered LSC. ACKNOWLEDGMENT The authors thank Armour and Co., Kankakee, IL, for their generous gift of Aliquat-336.
LITERATURE CITED (1) Ross, H. H. Anel. Chem. 1966, 38, 414-420. (2) Ross, H. H. Anal. Chem. 1965, 37, 621-623. (3) . . Salarla. G. 6.: Rulfs. C. L.: Elvina. P. L. Anal. Chem. 1983, 35. 983-985. (4) Schram, Eric "Organic Sclntlllatlon Detectors"; American Elsevier: New York, 1963; pp 21-28.
RECEIVED for review March 11,1981. Accepted May 18,1981.
Electronic Ruler for Digitizing Data S. R. Inman," S. A. Sibbald, and R. B. McComb Clinical Chemistry Laboratoty, Department of Pathology, Hartford Hospital, Hartford, Connecticut 06 1 15
A major advantage of high-performance liquid chromatography (HPLC) is that niultiple components of interest are analyzed within any one run. Under ideal conditions, each component is identified by the retention time of its associated chromatographic peak andl is quantitated by the peak height or area of this same peak. However, as the number of components increase, the qutintification becomes tedious and significantly adds to the total analysis time. The use of on-line integrators coupled with automated data processing apparatus
is one solution to this problem. An alternate solution, and one which is being used successfully in our laboratory, involves measuring the peak heights of the pertinent chromatographic peaks by means of a caliper coupled to a slide wire potentiometer and feeding the digitized signal to a small programmable calculator. All data reduction is performed by the calculator, and the results are printed on a tape. Our HPLC system is being used for the analysis of five antiepileptic drugs (AED).
0003-2700/81/0353-1723$01.25/00 1981 Amerlcan Chemlcal Soclety
1724
ANALYTICAL
CHEMISTRY, VOL. 53. NO, 11. SEPTEMBER 1981
1
2
3
4
5
MINUTES
Flgure 1. Chromatogram of anttapileptic drugs from a serum based standard (1) primidone. (2) ethosuximide. (3)phenobarbital. (4) 4nttrophenol (Internal standard). ( 5 ) phenytoin. (6) carbarnrepine. The mobile phase is 0.01 m l 1 L potassium phosphate buffer. pH 6.35. acetonttrile, 60140. The column system used was a 6 mm 1.d. C,, Radial Pak A carkidge held h a RCMIOO Radial compresskm module. Waters Associates. Mllford. MA. Flow rate is 2.5 mLlmin. DBtectoT wavelength Is 205 nm.
I
131.
CSM
561.
CAB\
Ethosvximide Phenobarbital
112. 29.844 106.
Phenytoh 14.919 180.
ckbmezepine
22. 6.926
**.
***
dsm
*.
um
*.. f
f
f
Fgue 4. Calculator printout of AEO resuils frmFlgue 1. Note that the m k hekht fa the hemal standard iS Cdnted lust before the resu(ts fw Rst &g (primldone). The softwa& wnvmbn of peak height in millivolts (GSBA value) to peak height in millimeters allows an independent check of these peak helghts wRh a w n v e n ~ruler. l The printing of me mlllivon signal is optional.
the
Figure 2. The electronic ruler.
EXPERIMENTAL SECTION Materials and Methods. The AEDs are analyzed by HPLC by a method similar to that of Soldin (1,2). (See Figure 1.) The programmable calculator and interface accessory is a Model HP97S, purchased from Hewlett-Packard. Corvallis, OR. The electronic ruler (Figures 2 and 3) was manufactured in-house as described below. The ruler is constructed of a IO-in. linear slide-wire (salvaged from a Honeywell Electronik 15 recorder, slide-wire part no. 78974-5, contactor part no. 369746-1 Honeywell, Philadelphia, PA). A 13 x 25 X 45 mm block made of PVC, on which are mounted the slidewire contactor and a pointer, is suspended above the side-wire by means of a 3/le in, diameter stainless steel rod. The rod passes through the center of the block allowing the contactor and pointer to easily traverse the length of the slidewire. The rod and contactor assembly are mounted securely to the slide-wire bracket. Also mounted on this bracket are a fixed pointer and a push button switch. The potential for the slide-wire is supplied by the stable reference voltage of the Analog to Digital converter module (Intersil 7101-8052). The voltage analog of position or peak height is converted to a three-digit BCD output by the ADC module upon activation of the push button switch mounted on the slide-wire bracket. Immediately following the conversion cycle (cycle time approximately 300 ms) the ADC module outputs a pulse that signals the HP97S to read the BCD data at the interface. The
data are then automatically processed by the software routine programmed into the calculator. The HP97S performs all required computations using the digitized voltage signals coming from the ruler since voltage is proportional to length. However, for purposes of manually checking results with a conventionalruler, we have incorporated a softwareconversion of voltage to length. This allows each peak height to be printed in millivolts and millimeters along with the drug concentration; see Figure 4. Calculations. Memory limitationsof the HP97S requires that the program for data processing of AED results (available on request from the authors) be stored on two cards. The data processing operation is started by insertion of card 1dter which the peak heights of the standards are entered from the electronic ruler. Program card 2 is then inserted into the calculator followed by entry of the peak heights of the unknown drugs. The peak height of the internal standard is always entered first, and the drug peak heights are entered in order of increasing retention time. To use the electronic ruler, the fixed pointer of the ruler is placed on the estimated peak base line and the sliding pointer moved to the peak maximum. This voltage is transmitted to the calculator by depressing the start conversion button. If a peak is missing, the sliding pointer is returned to the “zero” position and the button depressed, thereby transmitting a voltage equivalentto 0 mm for the peak height of the missing drug. After each entry, the peak height and correspondingdrug concentration is printed. Each set of results is grouped on the tape so that a triple space separates patients (see Figure 4).
RESULTS AND DISCUSSION
To estimate the measurement precision of this device, the heights of peaks from each of three AED chromatogrm were measured by six individuals. The heights of these 18 peaks were then remeasured by the same individuals using a conventional ruler. Precision, expressed as i1 standard deviation from the mean, averaged 0.4 mm for the electronic ruler and
1725
Anal. Chem. 1981, 53, 1725-1726
0.3 mm for the conventionalruler. As expected, precision was poorest for those peaks measured from the sloping base line with the highest standard deviation value being 0.8 mm for one set of primadone readings. This measurement precision was deemed satisfactory. The average bias between the two measurements was +0.2 mm. CONCLUSIONS The electronic ruler is easy to use, is mechanically reliable,
and has significantly reduced time spent in AED data reduction in our laboratory over the past 6 months.
LITERATURE CITED ( I ) Soldln, S. J.; Hill, J. G. Ciin. Chem. 1976, 22, 856-859. (2) Soldin, S. J. Clin. Biochem. 1880 73, 99.
RECEIVED for review March 9,1981. Accepted April 4, 1981.
Reactions during the Titration of Ammonia with Hypoclhlorite in the Presence of Bromide Alan H. B. Wu' and Howard V. Malmstadt" School of Chemical Sciences, Department of Chemistry, University of Illinois, Urbana, Mnois 6 180 1
There have been many titration investigations for determining ammonia with hypochlorite titrant in the presence of bromide (I-7),and these have indicated that hypochlorite reacts with bromide to form hypobromite that reacts subsequently with ammonia according to eq 1 and 2. However,
Br2NH3
+ - + +
+ OC1-
+ 30Br-'
OBr- GI\Nz 3HzO 3Br-
(1) (2)
it has been found in the present investigation that reactions 1 and 2 occur only at pH values higher than those normally used for the titration. For example, by monitoring the titration in solutions buffered a t p€I 10.5, reactions 1 and 2 are shown to proceed as reported. However, by monitoring the titration a t the more typical pH 8,6 with a photodiode array rapidscanning spectrophotometric titrator @), whereby spectral regions can be monitored without titration interruption, it is apparent that bromine is the reactant with ammonia. Data are presented to show that in a solution buffered a t pH 8.6, hypochlorite reacts with bromide to form bromine that reacts with ammonia in the titration procedure, according to eq 3 and 4.
+ HzO + OCl6 0 H - + 2NH3 + 31312' 2Br-
--
Br2 + C1- + 20H-
(3)
N2 + 6H2Q + 6Br-
(4)
Under a third set of conlditions, the titration of ammonia with hypochlorite and bromide in a lightly buffered solution at pH 8.6, a combination of reactions 1-4 are shown to occur. The reasons for this are discussed and are based on both spectrophotometric and pH data.
EXPERIMENTAL SECTION Reagents. The 0.05 M hypochlorite solutions are prepared from commercial bleach (the Chlorox Co., Oakland, CA). Standard ammonia solutions are prepared from primary ammonium sulfate (Mallinckrodt Chemical Works, St. Louis, MO). The 5 M sodium bromide solutions are also prepared in the usual manner. Two borate buffer solutions are prepared at pH 8.6. One lightly buffered solution is prepared by dissolving 7.51 g of sodium tetraborate decahydrate (Mallinckrodt Chemicd Works) in 1 L of water, and another solution is heavily buffered by dissolving 38.1 g of borate in 1 L of water. Both are adjusted to pH 8.6 with perchloric acid, A third buffer is prepared at pH 10.5 by dissolving 6.6 g of borate in 1 L of water and adjusting the pH to 10.5 by adding sodium hydroxide. Present address: Clinical Chemistry Laboratory, Hartford Hospital, CT 06115. 0003-2700/81/O353-1725$0 1.25/0
Procedure. Titrants are standardized by using 0.10 N arsenic trioxide (Mallinckrodt). Titrations proceed using 10 mL of buffer, 1 mL of 5 Pvl sodium bromide, 4 mL of water, and 5 mL of standard ammonium sulfate solutions varying in concentrations from 7 to 50 MM.
RESULTS AND DISCUSSION The titration of ammonia with hypochlorite in the presence of bromide zit pH 10.5 produces the curve shown in Figure 1. The reaction is monitored at 330 nm, the absorbance maximum for hypobromite. Figure 2 shows spectral scans from 240 to 3\50nm taken during various times of the titration of Figure 1 where these scans were taken. No traces of bromine, which has an absorption maximum of 267 nm, can be seen for this titration at pH 10.5. The apparent absorption before the end point of Figure 1 corresponds largely to the scattering of light by the microbubbles of nitrogen formed during the reaction shown in eq 2. In separate studies where the delivery of titrant is stopped at several times before the end point, it i3 noted that absorbance of the solution gradually returns to the base line as the last visible nitrogen escapes from solution. The titration of ammonia with hypochlorite in the presence of bromide that is heavily buffered at pH 8.6 produces the titration curve shown in Figure 3. The reaction is monitored at 267 nm, the absorbance maximum for bromine. Before the end point, the apparent absorbance is partly due to the scattering of light from nitrogen and partly to a trace of unreacted bromine produced during the continuous titration. Because of the high molar absorptivity of bromine in solution, trace amounte,can be spectrophotometrically detected. The presence of bromine was confirmed by the photodiode array titrator, where the absorbance curves are obtained at the various numbered points during the titration as shown in Figure 4. Curve 1 shows the presence of residual bromine before the end point. The presence of a large absorption band, shown in curve 2, is from excess bromine produced from reaction, eq 3, slightly after the end point, after all of the ammonia has bleen titrated. The high absorptivity of bromine produces the very sharp end point exhibited in Figure 3, distinctly different from the end point of Figure 1. The abrupt leveling off of absorbance is caused by an instrumental deviation from Beer's law. No traces of hypobromite can be seen in the curves of Figure 4. The titration of ammonia with hypochloritein the presence of bromide in a solution lightly buffered at pH 8.6 produces the titration curve shown in Figure 5A. Before the end point, the reactions summarized in eq 3 and 4 occur as confirmed 0 1981 American Chemlcal Society