suggesting that loss of boron to the decomposition products of silicate samples is insignificant. The effect of mannitol on the color development is slight. The absorbances of four aliquots of the 5-pg standard solution having no mannitol aiid the equivalent of 1 , 2, and 4 mg of mannitol per ml, respectively, were determined (Table IV). The boron-mannitol complex is apparently decomposed by the sulfuric acid. The color development characteristics of the carmine reagent apparently vary with the brand used. In the present study the intensity of the color reaches a maximum after 40 minutes and thereafter decreases: the fading rate is approximately 0.003, 0.005, aiid 0.010 absorbance units per hour for the 3-, 5, and IO-pg standards, respectively. This effect was negated by determining the absorbance of each solution 40 minutes after the addition of the carmine reagent. The absorbances of similar :samples and standards varied slightly from day to day, possibly influenced by variations in labora-
tory conditions. For this reason it was found advisable to run three standards with each batch of determinations. T o determine the precision of the method, 25 determinations of NBS 77 were made over a period of 2 weeks. These gave a mean of 137 ppm of boron, a standard deviation of 1.43, and a relative standard deviation of 1.04 %. The accuracy of any method of trace element analysis is difficult t o assess. However, the results of the recovery determinations are helpful in this respect: these suggest the accuracy is at best zt2z. In addition, several standard materials, on which boron determinations have been made by other workers, were obtained and analyzed. The results are compared in Table V. The present results compare quite favorably.
RECEIVED for review September 1, 1966. Accepted October 31,1966. This work was undertaken with the aid of a N.R.C. Operating Grant.
Optimum Conditions and Variability in Use of Pulsed Voltage in Gas-Chromatographic Determination of Parts-Per-MilIion Qua ntities of Nitrogen Dioxide M. E. Morrison and W. H. Corcoran Chemical Engineering ikboratory, California Institute of Technology, Pasadena, Calif.
IONIZATION DETECTOR:; (1-6) have recently been developed which have a very high sensitivity to specific compounds. In particular, the electron-capture detector (7) has been shown to be very sensitive to compounds with high affinities for free electrons. Because of the relatively high electronegativity of the nitrogen oxides, ari electron-capture detector was studied for its applicability in the detection of parts-per-million quantities of nitrogen dioxide. Earlier work (8) with an opposedflow detector operated in the electron-capture mode showed that the method could be applied to the quantitative analysis of nitrogen dioxide becween 5 and 150 ppm (9). Because of the desire to analyze :‘or nitrogen dioxide at concentrations below 1 ppm a plane-parallel (7) electron-capture detector was designed and built. The effects of temperature, flow rate, size of tritium source. voltage, and the means of applying voltage to the plane-parallel detector were studied. In the study of the methods for applying voltage, direct current and pulse modes were used. Absolute sensitivity is dependent upon the inherent noise from random emission of the @-raysfrom the tritium source. (1) R.D.Condon, ANAL.. CHEM., 31,1717 (1959). (2) 0.L. Hollis, Zbid.,3:3, 353 (1961). (3) A. Karman, L.GinRrida, and R. L. Bowman, J. Chromatog., 9, 1 3 (1962). (4) J. E. Lovelock, ANAL.CHEM., 33, 162 (1961). (5) J. E.Lovelock, J. Chromatog., 1, 35 (1958). (6) J. E. Lovelock, Nutu.ve, 182, 1663 (1958). (7) J. E.Lovelock, ANAL.CHEM., 35,474 (1963). (8) M.E. Morrison, R. (3. Rinker, and W. H. Corcoran, Ibid., 36, 2256 (1964). (9) The term ppm refers to parts-per-million and is defined here as mole fraction X lo6.
The magnitude of the noise can be calculated in an approximate manner from the Shot equation (IO):
-
i 2 = 2e’(a
+ l)ZoB,
In Equation 1, e’ is the charge of an electron, Io the quantity of @-rays emitted from the tritium foil per unit time, a the number of electrons formed by ionization per emitted @-ray, B, the electrometer bandwidth, and iy the mean square fluctuating current due to random arrival of electrons at the anode. Use of this equation gave a noise level of 6(a 1) x 10-13 amp. Therefore, the theoretical detection limit of the cell could be defined as that quantity of electron-capturing material which would yield a change in current of 6(a 1) x 10-13 amp. A value for a of approximately 5 was experimentally determined, and therefore the theoretical noise level was 3 X 10-’2 amp. Previous work (8) with the o p posed-flow detector resulted in a noise level of 1.6 X 10-11 amp. By means of an oscilloscope, measurements were made on the noise under normal operating conditions in the opposed-flow detector circuit. The noise seemed somewhat cyclic (10 cps) and was affected quite markedly by small changes in the flow of scavenger gas. The electrical noise was measured by replacing the detector with a 5 X 107 ohm amp. Thereresistor and was found to be less than 2 X fore the conclusion was that the high noise level was probably due to the flow pattern in the cell. In order to minimize the turbulence in the cell and hopefully to decrease the noise, the plane-parallel became of special interest.
+
+
~~
(10) K. R. Spangenberg, “Vacuum Tubes,” p. 306, McGraw-Hill, New York, 1948. VOL. 39, NO. 2, FEBRUARY 1967
255
3 0
/-A
-
DETECTOR TEMPERATURE 0 A
200OC
0
116°C 89OC
d
27OC
I
-
I
CONCENTRATION OF NO2, Bp.m. (@ 27OC,760 m m Hg)
Figure 1. Response of plane-parallel detector as a function of nitrogen dioxide concentration in samples containing nitrogen and up to 25% oxygen with temperature as a parameter and operation in the pulse mode at optimum conditions EXPERIMENTAL
Apparatus. A Loenco 15A gas chromatograph with a Loenco 15B electrometer was used in the study. Earlier work with that system was carried out with an opposed-flow Barber-Colman Model A-5042 electron-capture detector which had a 220 mc tritium source. The plane-parallel detector which was built was similar to one described by Lovelock (7). A source of approximately 180 mc of tritium obtained from the Radiation Research Corp. was used. The body of the detector was made from 11]4in. Teflon bar stock and 304 stainless steel. A diffusion screen of 150 mesh, 304 stainless steel with 0.0041-in. openings was used to minimize eddy effects. In operation, high-purity, dry nitrogen and commercial-grade oxygen, both obtained from Linde, were the diluent gases. They were dried by passage through a 4-fOOt length of 3]le-in.i.d. copper tubing packed with 13-X molecular sieve and mounted in a dry ice-acetone bath. Packing for the chromatography column used for separation of nitrogen dioxide from nitrogen, nitric oxide, and oxygen consisted of 10% by weight of SF-96 (a methyl silicone oil) on 40180 mesh Fluoropak 80. It was placed in a 20-foot length of 1I8-in., 304-stainless-steel tubing which had a wall thickness of 0.016-in. Details of the preparation of the column have been discussed (8, ZI). A Loenco 208 sampling valve made of 316 stainless steel was used to introduce the gas sample into the column. Nitrogen Dioxide. Nitrogen dioxide at a concentration of 1627 ppm, with a standard deviation of 70 ppm, was available in nitrogen, and samples were prepared as previously described (8). Procedure. The conditioned Fluoropak 80 column was operated at 22” C. Subsequent to the work reported earlier (8),in which nitrogen was used as a carrier and scavenger gas, argon was found to increase the sensitivity of the BarberColman A-5042 detector to nitrogen dioxide by a factor of (11) M. E. Morrison, “11. The Quantitative Determination of
Parts-per-Million Quantities of Nitrogen Dioxide in Nitrogen, Oxygen, and up to 75 ppm of Nitric Oxide by Electron-Capture Detection in Gas Chromatography,” Ph.D. Thesis, California Institute of Technology (1965). 256
ANALYTICAL CHEMISTRY
---$y:a ----L
\ I MIN.
I
0
Figure 2. Typical chromatogram for nitrogen dioxide in the presence of nitrogen, oxygen, and nitric oxide for the planeparallel detector under optimum conditions: 10 cc/min of carrier argon, 30 cc/min of scavenger argon, 50-v pulse potential, 150 psec period, 4 psec on-time, 200’ C detector temperature two. Therefore, argon was used as both a scavenger and carrier gas. In the study of the optimum conditions of operation, 0.5-cc samples of nitrogen containing 88.3 ppm of nitrogen dioxide were used. The purpose of the experiments was to determine the effect of the geometry of the plane-parallel detector on electron-capture detection and to compare the dc and pulse modes of operation. RESULTS AND DISCUSSION
For the d.c. method of operation with the plane-parallel detector, the most sensitive response was with a carrier flow of 10 cc per min of argon, a scavenger flow of 10 cc per min, and a detector potential of 4.5 v. The plane-parallel design required a lower potential and scavenger flow for optimum response in comparison to the opposed-flow Barber-Colman detector-Le. 4.5 us. 33 v and 10 cs. 85 cc per min., respectively. With the pulse mode of operation, the response of the planeparallel detector was relatively independent of voltage between 10 and 50 v. Below 10 v, odd peaks and anomalous results were obtained. The two important variables appeared to be fraction of on-time and total flow rate of argon. For the case of a carrier flow of 10 cc per rnin, a scavenger flow of 30 cc per min of argon and a fraction on-time of 0.027 gave the most sensitive response within the limits of experimental error. The effect of temperature on the response of the electroncapture detector when operated in the pulse mode was determined. The data are presented in Figure 1. It was found
o*--------l
1
07t
Table I. Calibration Data Collected for 1.5 Months for a Plane-Parallel Detector Operated in the Pulse Mode (11-2-64
to 12-18-64)" Nitrogen dioxide concn,
Fraction Fraction decrease in decrease in C'h-02, background C'NO~, background (PpmY currentC (PPmY currentc 0.2241 66.96 0.5997 20.72 0.5836 17.30 0,2053 65.21 0.5785 15.73 0.1623 60.06 0.1530 0.5384 14.29 57.98 0.1180 54.55 0.5350 11.60 0.5149 9.80 0.08998 47.53 0.07471 0.4351 9.55 40.16 0.4173 9.34 0.07993 36.04 0.4094 9.01 0.06963 35.36 30.42 0.3299 6.07 0.03896 0.3673 6.05 0.04274 29.51 0,03317 0.2804 5.82 25.30 23.02 0,2707 5.02 0.02762 0.2566 3.55 0.01535 21.50 On-time, 4 psec; period, 150 psec; argon carrier, 10 cc per min; argon scavenger, 30 cc per rnin; potential, 50 v ; and detector temperature, 200°C. * C ' S O =~ C~o,(P/760)(299.62/T), where C'NO?is the nitrogen dioxide concentration of the sample corrected from P to 760 mm of Hg and from Tto 299.62" K with the assumption of a perfect gas. Background current, (5.75 f 0.15) X amp. (I
0
\
&
?
!20
&
30 .
-
-40
1
50
60
70
NITROGEN DIOXIDE CONCENlAATION, p p m ( 0 5 c c SAMPLE @ 2646*C,760mmHg)
Figure 3. Calibration curve for nitrogen dioxide with the plane-parallel detector operated with optimum conditions for analysis noted in Figure 2 that the sensitivity was increased approximately fourfold in the concentration region from 1 t o 10 ppm of nitrogen dioxide when the temperature was decreased from 200 " to 25 " C, but the effect was not linear with temperature. In fact, nearly all of the increase in sensitivity was obtained in the range from 90" t o 25" C. This is the exact opposite of the observation by Landowne and Lipsky (12) for sec-butyl bromide. The effect of temperature o n the sensitivity of the detector is a characteristic of the substance t o be detected and a function of the electron affinity of the molecule and of the heat of dissociation for the negatively charged molecule to form a free radical and negative ion. By use of tritium foils with different activities per unit area, the effect of the size of the tritium-source on the response of the detector was determined. An increase in the strength of the tritium source did ircrease the response for a given quantity of nitrogen dioxide but the noise level was increased in the same proportion. Thus, the absolute sensitivity was not a function of source strength as long as a reasonable background current could be obtained. The acceptable size of source was in the range of 150 to 3130 mc of tritium. With this quantity of tritium, the value of u, the number of electrons collected a t the anode per emitted P-ray, was 5. Note should be made that the sensitivity tests on the plane-parallel detector were conducted using a tritium source of 150 mc. In the calibration studies, the source was 200 mc. When a pulsed voltage was used, the conditions for optimum sensitivity to nitrogen dioxide for a flow rate of carrier gas of 10 cc per min were found t o be: a scavenger flow rate of argon of 30 cc per min, a pulse output of 50 v with an on-time of 4 psec and a period of 150 psec, and a detector temperature of 200" C. For these conditions the noise level of the detector amp with a backwith a 150-mc tritium source was 4 X ground current of 2.6 X amp. The noise level is a (12) R. A. Landowne arid S. R. Lipsky, ANAL. CHEM.,34, 726 (1962).
quarter of that obtained with the Barber-Colman A-5042 detector operated in the d.c. mode and half of that with the planeparallel unit in the d.c. mode. The detector temperature of 200" C was chosen over 25" C for two reasons: the calibration curve at 25 " C was much more nonlinear than a t 200 O C, and less adsorption of column bleed on the tritium foil was experienced a t 200" C. A chromatogram for the optimum conditions of analysis in the pulse mode is shown in Figure 2. Nitrogen was the diluent. The concentration of nitrogen dioxide was 4.22 ppm and that of oxygen was 18.15 X l o 4 ppm. The sample pressure and temperature were 748.7 mm of mercury and 26.2" C, respectively. Oxygen was present just to check separation, and the multiple oxygen peak was developed merely by decreasing the attenuation by factors of two from a value of 128 t o 2 during the course of the oxygen peak. The sample volume was 0.5 cc. Calibration data were obtained for nitrogen dioxide when the detector was operated under optimum conditions with pulsed voltage. The data are tabulated in Table I and presented in Figure 3. They are plotted with the fraction decrease in current as a function of the nitrogen dioxide concentration in order t o compensate for day-to-day variations in the background current. For the best curve which could be drawn through the data, the average per cent deviation was 3.2 from 3 to 25 ppm of nitrogen dioxide and 3.4 from 3 to 75 ppm. The standard deviation was 0.59 ppm from 3 to 25 ppm and 1.32 ppm from 3 to 75 ppm. All concentrations of nitrogen dioxide were corrected to 26.46" C and 760 mm of mercury as shown in Table I. The reason for the correction was to compensate for slightly different sample pressures and temperatures. The fit of the calibration data to the theoretical equation proposed by Lovelock (7) was again unacceptable for this work. The chromatographic determinations compare very favorably with chemical techniques for the analysis of nitrogen dioxide. Chemical methods used for the analysis of nitrogen VOL 39, NO. 2, FEBRUARY 1967
257
dioxide for the range of concentrations of 1 to 75 ppm are the phenol-disulfonic acid technique (ASTM D 1608-60) and the Saltzmann technique (13). The phenol-disulfonic acid method requires 1-liter samples and has given in this laboratory an average absolute deviation of 6 % and a standard deviation of 3 ppm. The Saltzmann procedure requires 250-cc samples. An average absolute deviation of 3-4 % has been reported (13) in the concentration range from 8 t o 45 ppm. The phenoldisulfonic acid method is used for the determination of the total nitrogen oxides as nitrogen dioxide, whereas the Saltzmann procedure is used for the analysis of only nitrogen dioxide in the presence of any other nitrogen oxides. Because (13) B. E. Saltzmann, ANAL.CHEM.,26, 1949 (1954).
the analysis time is 5 min. for gas chromatography cs. 3 to 4 days with the phenol-disulfonic acid method and the sample size is 0.5 cc us. nominally 1 liter, the analysis by gas chromatography shows advantages for the conditions studied. Sample size would be the main advantage relative to the Saltzmann procedure.
for review May 11, 1966. Accepted September 26, 1966. The research project was financed by the Division of ~i~ pollution, Bureau of State Services, United States public Health Service. Fellowship support for M, E. Morrison was contributed by the Dow Chemic; Co. and the Union Carbide Corp.
RECEIVED
Gravimetric Determination of Mercury, Lead, and Platinum Using Trirnethylphenylammonium Iodide W. W. White and J. R. Zuber Electronic Componenfs and Decices, Radio Corp. of America, Someroille, N. J. 08876
RAPIDSEMIMICRO METHODS are described for the gravimetric determination of Hg(II), Pb(II), and Pt(1V) in quantities of 4 t o 60 mg in various complex matrices by use of trimethylphenyiammonium iodide (TMPI). This reagent was used first by Pass and Ward (1) t o determine cadmium in the presence of zinc. The reagent was utilized further by Burkhalter and Solarek ( 2 ) for the gravimetric determination of bismuth and by White and Zuber (3) for the gravimetric determination of gold. Other organic reagents have been reported to precipitate the halide complexes of mercury and platinum but each reagent has its own degree of selectivity (4). Ethylenediamine or propylenediamine, each in the presence of copper sulfate, precipitates the Hg14-2 complex (5) whereas cinchonine, quinine, or hexamine each precipitates with iodoplatinic acid (6). Tetraphenylarsonium bromide has been used to give a precipitate with the PtBr8-z complex (7). Relatively few organic compounds, however, have been reported for precipitating platinum, as indicated by Beamish in his recent review of gravimetric methods for the determination of the noble metals (8). Picrolonic acid (9), sodium anthranilate ( I O ) , and thionalide (11) are used to determine milligram quantities of lead, but (1) A. Pass and A. M. Ward, Analyst, 58,667 (1933). (2) T. S. Burkhalter and J. F. Solarek, ANAL.CHEM.,25, 1125-6 (1953). (3) W. W. White and J. R. Zuber, Ibid..36, 2363-4 (1964). (4) K. Kodama, “Methods of Quantitative Inorganic Analysis,” Interscience, New York, 1963. (5) Ibid.,p. 153.
(6) S . Takagi and Y. Nagase, J. Pharm. SOC.Japan, 58, 60-6 (1938). (7) H. Bode, 2. Anal. Chem., 133, 95 (1951). (8) F. E. Beamish, Talanta, 13, 773-801 (1966). (9) H. Imai, J. Chem. SOC.Japan, 76, 770 (1955). (10) J. F. Welcher, “Organic Analytical Reagents,” Vol. 11, p. 200, Van Nostrand, New York, 1947. (11) K. Kodama, “Methods of Quantitative Inorganic Analysis,” p. 160, Interscience, New York, 1963. 258
ANALYTICAL CHEMISTRY
many cations interfere. The precipitation of lead as the sulfate from dilute sulfuric acid solutions is not entirely quantitative (12). The addition of alcohol reduces the solubility but a t the same time increases contamination by elements such as bismuth, calcium, and silver. The electrodeposition of lead as the dioxide is a n accurate method; however, arsenic, chloride, cobalt, manganese, and many other elements interfere (13). EXPERIMENTAL
Reagents. The precipitating solution for mercury and lead is prepared by dissolving 60 grams of potassium iodide and 25 grams of trimethylphenylammonium iodide (Eastman Organic Chemicals, reagent 4423) in 1 liter of water. The mercury wash solution is prepared by diluting 1 part of the precipitating solution with 4 parts of water. The solution is stable for several weeks. The lead wash solution is prepared by acidifying the above wash solution with hydrochloric acid to give a 2x acid solution by volume. This solution remains stable for only a few hours. The toluene wash solution is prepared by adding 25 ml of ethanol to 1 liter of toluene. The buffer and complexing solution is prepared by dissolving 150 grams of sodium citrate dihydrate and 200 grams of anhydrous sodium acetate in warm water and diluting to 1 liter. The mercury and lead standard solutions are prepared by weighing 1.0000 gram of each metal, dissolving carefully in dilute nitric acid, and diluting separately to 0.5 liter with distilled water. The platinum standard solution is prepared by weighing 1.0000 gram of the metal, dissolving in aqua regia, and diluting to 0.5 liter with water. (12) W. F. Hillebrand, G. E. F. Lundell, H. A. Bright, and J. I. Hoffman, “Applied Inorganic Analysis,” 2nd ed., p. 227, Wiley,
New York, 1953. (13) K. Kodama, “Methods of Quantitative Inorganic Analysis,” p. 158, Interscience, New York, 1963.
