1196
Anal. Chem. 1984, 56, 1196-1198 100 r
Table I.
Area R a t i o s of 4-Cl-2,6-DMP/4-Br-2,6-DMP
L
solution no. 1
6
t
/'
2501 k u
2 3 mean i std dev
,'
n.
V
I 25
1
1
75
1
I
I25
I
l
1 75
l
1
2 25
PPM CHLORINE IN TEST ATMOSPHERE
Figure 3. Plot of concentration in ppm Cl, ( x axls) vs. % recovered (y axis) by using a constant 10-min collection time. Each point rep-
resents the mean of three separate analyses. Transformation of these data to a log-log plot by using the actual amount of CI, (pg) recovered gives a linear regression with the followlng equation: log Y = 0.58 16 log X .
+
of 4-C1-2,6-DMP formed would not be directly proportional to the amount of Clz in the atmosphere but rather would also depend upon the rates of formation of the various chlorinating species. In addition, the conditions of low pH and the presence of oxygen may cause competing side reactions to occur such as formation of quinones from 2,6-DMP and 4-C1-2,6-DMP
20 p g of NH, added
con tro1 (without NH,)
0.095 0.094
0.095 0.096
0.1048
0.097
i
0.0058
0.1047
0.098
*
0.0053
studies (2). As can be seen in Table I, less than a 2% difference is observed in the 4-C1-2,6-DMP to 4-Br-2,6-DMP ratios for the spiked and unspiked samples. This is to be expected because of the large molar ratio (200 to 1) of 2,6DMP to NH3 as well as the low pH (2-3) of the solution which is well below that which favors the formation of NHzC1(14). Chromatographic analyses of Clz in the presence of NH3 and chloramines from gaseous samples reported here yielded results comparable to that of Ellis and Brown (8)from residual chlorine in wastewater. The use of fused silica capillary columns virtually eliminated interferences due to coeluting impurities from extraction solvents. Kinetic studies showed that hydrolysis of chloramines, specifically monochloramine, was sufficiently slow to be negligible. The sensitivity of the method is sufficient for determination of free Cl, in a 0.1 ppm Clz test atmosphere using a collection time of 10 min and a sampling rate of 0.73 L/min. With its specificity for free Clz, this methodology should be a valuable complement to continuous analyzers that measure total chlorine. Registry No. Clz,7782-50-5; NH3, 7664-41-7;2,6-DMP, 576-
(12).
The predominant chloramine found at the Clz and NH3 concentrations that apply in toxicity studies (ca. 1ppm) will be monochloramine (NHzC1) (13). Monochloramine hydrolyzes in aqueous solution to produce NH3 and HOC1. The presence of HOC1 from any other source besides free Clzwould cause a positive interference in the free Clz determination. NH2Cl was added to a solution containing 2,6-DMP and internal standard. The amount of 4-C1-2,6-DMPformed, which would be proportional to the amount of NHzCl that had decomposed, was measured with respect to time. The rate of disappearance of NH2Cl followed pseudo-first-order kinetics with a half-life of 23 h. Since the total number of moles of NHzCl to be formed cannot exceed that of the Clz, and the total sampling time including extraction is less than 30 min, the hydrolysis of NHzCl occurs much too slowly to interfere with the assay. The possibility that NH3 can interfere with the free Clz determination was investigated by spiking solutions of 2,6DMP with NH3, purging atmosphere containing 1 ppm Clz through the solutions for 10 min, and comparing the amounts of 4-C1-2,6-DMP formed with that of solutions which contained no NH3. Twenty micrograms of NH3 was chosen for the spiking experiment in order to mimic the higher concentrations of NH3 that could be encountered during toxicity
LITERATURE CITED Weedon, F. R.; Hartzell, A.; Setterstrom, C. Contrib. Boyce Thompson Inst. 1940, 1 1 365-385. Barrow, C. S.; Kociba, R. J.; Rampy, L. W.; Keys, D. G.; Albee, R. R. Toxicol. Appl. Pharmacol. 1979, 49, 77-88. Barrow, C. S.; Dodd, D. E. Toxicol. Appl. Pharmacol. 1979, 49, 89-95. Well, 1.; Morris, J. C. J . Am. Chem. SOC. 1949, 7 1 , 1664-1667. Standard Methods for the Examination of Water and Wastewater", 14th ed.; American Public Heath Assoclation: Washington, DC, 1976. Rune, H. 2. Anal. Chem. 1962, 789, 111-114. Rigdon, L. P.; Moody, J. G.; Frazer, J. W. Anal. Chem. 1878, 50, 465-469. Ellis, J.; Brown, P. L. Anal. Chim. Acta 1981, 124, 431-436. Corbett, R. E.; Metcalf, W. S.; Super, F. G. J . Chem. SOC. 1953, 1927-1929. Wong, G. T. F. WaterRes. 1980, 5 1 , 51-80. Swain, G. C.; Crist, D.R. J . Am. Chem. SOC. 1972, 9 4 , 3195-3200. Thompson, R. H. "The Chemistry of the Quininoid Compounds"; Patal, Saul, Ed.; Wiiey: New York, 1974; Chapter 3. Sisier, H. H.; Neth, F. T.; Drago, R. S.; Yaney, D., 1954, J . Am. Chem. SOC. 1954, 7 6 , 3906-3909. Morris, J. C. "Proceedings of the Conference on The Environmental Impact of Water Chlorlnatlon"; Joiiey, R., Ed.; NTIS: Springfield, VA, 1975; pp 27-41. I
RECEIVED for review September 12,1983. Accepted January 23, 1984. This research was presented in part at the 1983 Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy in Atlantic City, NJ.
Determination of Aluminum In Hemodialysis Concentrates by Electrothermal Atomic Absorption Spectrometry Pierre Allah,* Yves Mauras, and Francis Der Khatchadourian
Laboratoire de Pharrnacologie, Centre Hospitalier Universitaire, 49036 Angers Cedex, France In 1972, Alfrey (I) demonstrated the responsibility of aluminum in the occurrence of a progressive fatal neurological syndrome in patients treated by hemodialysis. It is recognized that the content of aluminum in the dialysis fluid plays a
major role in this intoxication and that its reduction below 10 pg/L can prevent its occurrence. Aluminum in the dialysis fluid usually results from water, but the possibility that the concentrate added to water to
0003-2700/84/0356-1196$01.50/00 1964 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 7, JUNE 1984
1197
1-
1
c
100
c +so
0.7
Ci50
c t20
W
V 0,5
0. Y
z a
m
Ct20
V
K
z a m
0 VI
C
a
C
E
0 v)
a
o , ~
Ir BLANK
(
0
L
Figure 1. Recorder traces of aluminum absorbance for blank, concentrate, and concentrates with addition of 20, 50, 100 pg/L of aluminum. Volume injected was 10 pL.
Figure 2. Recorder traces of aluminum absorbance for three replicate injections of blank, Concentrate,and concentrates with addition of 20 and 50 pg/L of aluminum. Volume injected in the furnace was 10 pL.
obtain the dialysis fluid could contain aluminum has been raised. Determination of aluminum by electrothermal atomic absorption spectrometry and inductively coupled plasma has been described in water, plasma, urine, and blood but not in the dialysis concentrate which has a very high level in minerals. A typical composition of the concentrate (in g/L) is as follows: NaC1, 220; MgCl2.6H20,5.3; CaC12.6H20,13; sodium acetate, 166; i.e., about 10-fold more concentrated than seawater. Determination of aluminum in the concentrate is difficult because after dilution the sensitivity is not sufficient and without dilution (direct injection of the concentrate in the furnace) the results obtained are erratic. Our purpose is to present an improved procedure for this assay.
RESULTS AND DISCUSSION Figure 1 shows recorder traces of absorbance in peak height mode for the blank and concentrates without and with additions using the HGA 500. The linearity of the response is correct and the sensitivity, expressed as the concentration giving 0.0044 absorbance units, was 0.8 pg/L. The sensitivity can be improved by injecting 15 pL in the furnace. Figure 2 shows recorder traces in peak height mode for three replicate injections of blank, concentrate, and two standards (+20, +50 pg/L). In this example, the reproducibility is good enough for such a difficult analysis. The coefficient of variation calculated from ten determinations on a concentrate with a mean value of 8 pg/L of aluminum was 19% (range of values 6-10 pg/L). The signal from the blank was due to traces of aluminum in Suprapur nitric acid from Merck, which was more contaminated than Normapur nitric acid from Prolabo. After these determinations, it was necessary to clean out the white deposits formed on the inner surfaces of the contact cylinders of the HGA 500. Similar good results in aluminum determination were obtained by using the Varian GTA 95 instrument with pyrolytic tubes, but the white deposits were on the internal surface of the funnel and of the shroud of the workhead center block and not on the electrodes. We have had the opportunity to measure aluminum in dialysis concentrate from different French manufacturers and we never observed a level above 35 pg/L in any concentrate and often the levels were below 10 Fg/L. For a few years, we have been looking for a reliable method of measuring aluminum concentrations in dialysis concentrates and we have never obtained reproducible results until we used nitric acid in very high concentration as a matrix modifier. Without nitric acid, it is necessary to dilute the concentrate more than 10-fold to obtain a linear response, but with a loss of sensitivity. We tested other nitric acid concentrations and we observed the best results with a final concentration of about 40%. This high level of nitric acid has a less deleterious effect
EXPERIMENTAL SECTION Apparatus. The instrument used was a graphite furnace atomizer HGA 500 with an autosampler AS 40 from Perkin-Elmer and a spectrophotometer Model 875 from Varian with a HP 85 microcomputer. The furnace operating parameters were as follows: drying phase, 100 "C for 20 s and 150 "C for 20 s in ramp mode; charring phase 1400 "C for 50 s in ramp mode; atomization 2500 "C, 0 ramp (max power) 2 s hold; clean out 2500 "C for 3 s. Argon gas flow was used during all steps except for atomization which was done in a stop flow condition. The light sources were an aluminum hollow cathode lamp and a deuterium arc for background correction. The wavelength used was 309 nm. Pyrolytic graphite tubes from Perkin-Elmer were used. A Varian 975 spectrometer with a GTA 95 furnace and a sample dispenser was also used. Reagents. The reagents used were concentrated nitric acid, with little aluminum, from Merck (Suprapur) or Prolabo (Normapur) and aluminum-free water obtained by deionization and reverse osmosis. The standard additions method was used, adding 20,50, and 100 pg/L of aluminum to the concentrate to be analyzed. The sample preparation was very simple: To 500 pL of blank (water) and concentrate in polyethylene microvials was added 300 pL of concentrated nitric acid. Ten or fifteen microliters was injected in the furnace with the autosampler.
1198
Anal. Chem. 1984, 56, 1198-1199
on the graphite tube than we had feared and it is possible to make more than 50 measures with the same tube. But the results are very dependent on the quality of the graphite tubes, and quality can vary in the same batch of pyrolytic tubes. Sometimes replacing peak height by peak area mode can improve reproducibility and linearity. The results obtained by electrothermal atomic absorption spectrometry are in good accordance with those obtained by inductively coupled plasma spectrometry using the conditions described in a previous paper (2), modified for a simultaneous spectrometer, JY 48, with background correction by automatic displacement of the entrance slit.
In conclusion, in our experience, addition of a high concentration of nitric acid dramatically improved aluminum determination in concentrates used to obtain dialysis fluid. Registry No. Aluminum, 7429-90-5; nitric acid, 7697-37-2. LITERATURE CITED (1) Alfrey, A. C.; Le Gendre, G. R.; Kaehny, W. D. New Engl. J . Med. 1076, 294, 4 , 184-188. ( 2 ) Allah, P.; Mauras, Y . Anal. Chem. 1070,51, 2089-2091.
RECEIVED for reivew October 12, 1983. Accepted February 9, 1984.
Dual Sample Injection for Gas Chromatographic Determination of Sulfur Species with Flame Photometric Detection Floyd A. Barbour,* Robert E. Cummings, and Frank D. Guffey University of Wyoming Research Corporation, P.O. Box 3395, University Station, Laramie, Wyoming 82071 The processing of oil shale whether by in situ or surface retorting generates gaseous sulfur products. During the development of near production-size processes, the release of these products needs to be monitored not only for compliance with air quality standards but also as a data base for input into the design or selection of a product gas cleanup system. The need for continuous monitoring demands that the system be fully automated and reliable. The use of a flame photometric detector (FPD) in gas chromatography (GC) lends itself nicely to the analysis of sulfur gases because the sensitivity for sulfur-containing compounds is in the subnanogram range. The detector has a major drawback in that the dynamic range of the detector only encompasses about 2 decades of sulfur concentrations, which is considerably less than the differences between the major sulfur species (hydrogen sulfide) and the minor species (carbonyl sulfide, methyl and ethyl mercaptan, carbon disulfide, and thiophene) observed in previous monitoring of oil shale retorts (I, 2). Gangwal and Wagner (3) overcame limitations imposed by the small dynamic range of the detector by evacuating the sample loop to an appropriate pressure. However, this technique requires precise vacuum control to evacuate to the same pressure each time and does not eliminate the problem of handling largely different component concentrations in the same gas sample. For this reason we connected two sample valves in series, one with a small sample loop for components present in high concentrations and one with a large sample loop to handle components in lower concentrations. Samples from both valves were sequentially injected during the same determination. EXPERIMENTAL SECTION A Hewlett-Packard 5840 gas chromatograph equipped with a single FPD and conventional six-port gas sampling injection valve was used for the analysis. Modification in the sample injection system was made in order to accommodate a small sample size as well as dual injection by adding a Rheodyne sample valve, type 50. This valve was modified as shown in Figure 1 by replacing the stator with one which contained no holes in the one- and four-port positions, which are normally used for the attachment of the sample loop. A sample loop was then scratched with a scalpel into the rotor side of the stator between the one and four pmitions. This produced a sample valve with a loop approximately
ROTOR
STATOR
Figure 1. Construction of 1-pL sample loop. SAMPLE IN
n
CARRIER OUT
Figure 2. Automatic gas sampling valves showing dual sampling capabilitles.
1pL in size that was attached in series to the conventional sample valve with approximately a 50-wL loop. The loops were connected, as depicted in Figure 2, so that the sample gas flowed through the larger sample loop and then to the smaller loop. The carrier gas was also plumbed into the two valves in series. Figure 2 shows the small sample valve in the injection position and the large sample valve loop in the fill position. The small sample was injected at the start of the determination and after 2 min the large sample was injected onto the column. The sample loop pressure and flow were controlled by using a water tower approximately 1 m in height. The inlet pressure was controlled to 5.6 kPa above atmospheric (approximately 78 kPa at Laramie, WY) allowing excess sample to bubble through the water and exit through a vent. The sample loops were held
This article not subject to US. Copyright. Publlshed 1984 by the American Chemical Society