380
Anal. Chem. 1988, 60,380-383
grows. Furthermore it is interesting to note that at higher intensities the sensor a t the cold tip also indicates a temperature increase, but less than the internal probe. Though the experiments described above are exemplary, they show clearly that when conductance cooling in a closed-cycle helium refrigerator is employed quite good temperature stability can be realized even with relatively high laser powers. One always has to be aware, however, of the possibility of a slight rise in effective temperature at high excitation intensities. This temperature increase is also observed when precautions have been taken to ensure good thermal contacts between sample and cold tip, as in our experimental setup. When thermal contacts are bad, like for tetracene in the polyethylene measurements where no attention was paid to the contact of the polymer and the sample holder, increased laser intensities give rise to poignant temperature effects which are fortunately easily recognizable.
LITERATURE CITED Hofstraat, J. W.; Gooijer, C.; Velthorst, N. H. I n Molecular Luminescence Spectroscopy: Methods and App//caflons,Part 2 ; Schulman, S . G., Ed.: Wiley: New York, to be published. Wehry, E. L.; Mamantov, G. I n Molecular Ruorescence Spectroscopy: Wehry. E. L., Ed.; Plenum: New York, 1981; Vol. 4, p 193. De Lima, C. G. CRC Crit. Rev. Anal. Chem. 1986, 76, 177. Vibrational Spectroscopy of Trapped Species: Hallam, H. E., Ed.: Wiley: New York, 1973. Hofstraat, J. W.; Freriks, I.L.; De Vreeze, M.; Gooijer, C.; Velthorst, N. H., submitted to the J. Phys. Chem. Hofstraat, J. W.; Schenkeveld. A. J.; Gooijer, C.; Velthorst, N. H., submitted for publication in Spectrochim. Acta. Cofino, W. P.; Van Dam, S. M.; Karnming, D. A.; Hoornweg, G. Ph.; Gooijer, C.; MacLean, C.; Velthorst, N. H. Mol. Phys. 1984, 57, 537. Hofstraat, J. W.; Schenkeveld, A. J.; Gooijer, C.; Velthorst, N. H., submitted for publication in J . Mol. Struct. Griesser, H. J.; Bramley, R. Chem. Phys. Lett. 1982, 88, 27. Hunadi, R. J.: Helmkamp, G. K. J . Org. Chem. 1978, 43. 1586.
RECEIVED for review May 29,1987. Accepted September 22, 1987.
Determlnation of Aluminum in Dialysate Concentrates by L'vov Platform Graphite Furnace Atomic Absorption Spectrometry Johanna Smeyers-Verbeke*
Farmaceutisch Instituut, Laboratory for Pharmaceutical and Biomedical Analysis, Vrije Universiteit Brussel, Laarbeeklaan 103, B-1090 Brussels, Belgium Dierik Verbeelen
Academisch Ziekenhuis, Renal Unit of the Department of Medicine, Vrije Universiteit Brussel, Laarbeeklaan 103, B-1090 Brussels, Belgium The determination of aluminum in biological materials by means of graphite furnace atomic absorption spectrometry has been the subject of many publications (1-5). Most attention has been given to the analysis of serum since serum A1 values can be used for diagnosing and for monitoring dialysis patients a t risk of aluminum intoxication (6, 7). In patients with renal failure treated by means of dialysis it is well established that, besides the ingestion of aluminum containing phosphate binders, aluminum contaminated water and dialysis fluids can cause aluminum toxicity. It is evident that proper control of A1 concentrations in water and dialysis fluids reduces the risk of an aluminum accumulation in these patients. Although the water used for the dilution of the dialysate concentrates seems to be the main source of the aluminum present in the dialysis fluids, a contamination of the concentrates themselves remains possible. Most work on dialysis solutions has been done on diluted which are prepared before use from hemodialysis fluids (8,9) dialysate concentrates by a dilution of about 35 times with water. The problem with the analysis of the concentrates is the very high salt content of these solutions. Typically they contain about 400 g/L of sodium, potassium, calcium, and magnesium chloride and sodium acetate. Here we report on the determination of aluminum in dialysate concentrates by means of graphite furnace atomic absorption spectrometry using the L'vov platform and ammonium nitrate as matrix modifier. The difficulties encountered during the development of the procedure are discussed. MATERIALS A N D METHODS Equipment. A Zeeman 3030 atomic absorption spectrometer equipped with an AS-60 autosampler and a PR-100 printer (Perkin-Elmer Corp., Norwalk, CT) was used for the measure-
ments. Pyrolytic graphite tubes (part no. B010-9322) with pyrolytic platforms (part no. B012-1091) were used. The instrumental conditions were as follows; drying at 160 "C for 15 s in ramp and 15 s in hold mode; charring at 600 "C for 30 s in ramp mode and 10 s in hold mode and 1600 "C for 40 s in ramp and 10 s in hold mode; atomization at 2500 "C, 0 s ramp (maximum power) 5 s hold; clean out at 2700 "C for 4 s. Argon gas flow was 300 mL/min except for the atomization, which was done in gas stop condition. An aluminum hollow cathode lamp was used at a wavelength of 309.3 nm. Background correction was used for all measurements. Contamination Control. Precautions were taken to avoid contamination as described elsewhere ( I ) . Composition of the Concentrate. The composition of the concentrate used in this investigation was as follows: NaCll87.0,
[email protected],KC19.5, CaC12.2Hz08.8, and MgC12.6H203.3 g/L. For use in the dialysis unit this solution was diluted 32 times with reverse osmosis purified water. Reagents. Standard Al solutions were prepared from Titrisol standard solutions containing 1 g/L of aluminum (Merck, Darmstadt, FRG). The water used to prepare all solutions was doubly distilled in a quartz device just before use. It contained no detectable Al. Nitric acid was of Suprapur grade and ammonium nitrate of "pro analysis" grade (Merck). A 100 g/L (10%)solution of ammonium nitrate in water was prepared in a plastic container and used as matrix modifier. Spiking of Concentrates with Al. The addition of A1 to concentrate solutions was performed by pipetting 20,50, and 100 pL of a 1 mg/L A1 solution to 1mL of concentrate. This results in samples with,respectively, 19.6,47.6, and 90.9 pg/L of Al added. For comparison, standard solutions were prepared in the same way by replacing 1 mL of concentrate by 1 mL of HN03 0.2%. Recommended Analytical Procedure. Standards of 0, 10, 20, 40 and 60 pg/L of aluminum in "OB 0.2% (v/v) were prepared in quartz volumetric flasks. These were used during
0003-2700/88/0360-0380$01.50/00 1988 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 60, NO. 4,FEBRUARY 15, 1988
one week. Standards prepared in HN03 1.5% (v/v) to match the acid concentration of the concentrate samples gave a similar absorbance. Since we now use standards prepared in HNOB0.2% (v/v) for serum Al determinations these standards were preferred for practical reasons. Samples were prepared in the autosampler cups by adding 750 p L of HN032% (v/v) to 250 pL of the concentrate. A blank was prepared in the same way by adding the nitric acid to 250 pL of water. The autosampler was programmed to inject 10 pL of standard or sample and 5 pL of the matrix modifier. The measurements were performed by using the peak area mode (integration time 5 s) with the instrumental conditons given above. For each sample the mean of three measurements was used. The absorbance (Ass) of the sample is corrected for the blank absorbance. The concentration of Al in the samples was calculated from the calibration line. The A1 found in the blank, which was always less than 3 pg/L (precision *4%), came predominantly from the nitric acid and not from the ammonium nitrate.
I
Inject i o n s
LO
m
loo
b
$i 0.030
20
LO
60
Injections
RESULTS AND DISCUSSION Choice of Tube. We started our investigation with a standard uncoated graphite tube. No matrix modifier was used. The optimal ashing temperature for the samples was found to be 1750 "C. At 1450 "C, which is the optimal temperature for standard Al solutions, no useful atomization signal was obtained. This was due to an extremely high background. At 1750 "C the background signal had almost completely disappeared but low recoveries of Al added to the concentrate were observed, which points to the severe matrix effects of the concentrate matrix. Moreover, the slope of the standard addition lines decreased and the reproducibility got worse with an increased number of injections. White deposits were observed a t the inner surfaces of the graphite cylinders. This ended up causing a bad contact between the graphite tube and the contact cyliners and thus it was impossible to perform any further analysis. Our experience with pyrolytic coated graphite tubes confirms the observations of Allain et al. (10) in the sense that bad reproducibility was obtained. Addition of high nitric acid concentrations as proposed by these authors was not evaluated since these dramatically shorten the lifetime of the tube. Moreover high acid concentrations must be avoided to maintain the blank absorbance at an acceptable level (1). The above considerations prompted us to evaluate the performance of the L'vov platform. Optimization of Instrumental Parameters. The optimal ashing temperature for standards and samples was found to be 1600 "C. The addition of the matrix modifier had no influence on this temperature. The first ramp ashing step during 30 s to 600 "C was found necessary to avoid sputtering during thermal pretreatment. No attempt was made to decrease the ramp time period of the second ashing step. Matrix Effects. As follows from Figure l a low recoveries are obtained for spiked concentrates that are diluted two times with water when compared with a standard solution with a similar concentration. Moreover a decrease in the absorption signal is observed with an increasing number of injections. This decrease in sensitivity was accompanied by an increase of the background signal. Fourfold dilutions with water or "OB 2 % had no effect on these interferences nor could a stable signal be obtained. A similar change of the sensitivity for the A1 standard solution was not found. Again salt deposits were observed at the ends of the graphite tube and at the inner surfaces of the contact cylinders. The A1 seemed to become incorporated in the matrix residues since after these measurements, injection of a dilute nitric acid solution initially resulted in a very high absorbance which only after a few injections returned to its base line value.
"
A
20
381
. . 0,060.
0,030 L 100
20G
1njec:Ions
Figure 1. Evaluation of the sensitivity with increased number of injections: (a)Concentrate spiked with AI 47.6 pglL; twofold dilution with water (0).Standard AI solution with similar concentration (0). (b) Concentrate spiked with AI 90.9 pglL; fourfold dilution with water and ammonium nitrate added (0).Standard AI solution with similar concentration (0). (c) Concentrate spiked with AI 90.9 pg/L; fourfold dilution with HNO, 2 % (v/v) and ammonium nitrate added.
Effect of Ammonium Nitrate. After the previous measurements the contact cylinders were cleaned out and a new graphite tube was installed. Several firings were then necessary to eliminate contamination resulting from the clean out procedure. The effect of ammonium nitrate on the concentrate diluted four times with water is shown in Figure lb. As can be seen the recovery is markedly improved but a 15% decrease in sensitivity is still observed after about 40 injections. Again an increase of the background absorption signal was then noticed. Inspection of the graphite tube and contact cylinders however revealed that deposita were not preceptable. This indicates that in the presence of ammonium nitrate a better elimination of the concentrate matrix is obtained. A stable signal was observed with concentrate samples that were diluted four times with "OB 2%. This is shown in Figure ICfrom which i t becomes evident that even after 200 injections an absorbance which is completely comparable with the initial signal is obtained. The absorbance ( A d of the background signal never exceeded 0.15 and was most often less than 0.10. Figure 2 reveals that the concentrate matrix alters the peak shape of the absorbance signal. A much sharper peak is seen and the A1 peak appears later in the concentrate matrix. Comparison of parts a and and b of Figure 2, where the absorbance signals for a standard solution and a concentrate, respectively, with and without the addition of ammonium nitrate is shown, indicates that the matrix modifier seems not to influence the absorption process of AI in either the standard or the concentrate. For both solutions the peak shape and the Ai appearance time look similar whether or not ammonium nitrate is added. Therefore the beneficial effect of ammonium nitrate in the determination of aluminum in dialysate concentrates is most probably due to a better elimination of the matrix by an increased volatility of the concomitants. Ediger et al. (12) were the first to propose ammonium nitrate to
382
ANALYTICAL CHEMISTRY, VOL. 60, NO. 4, FEBRUARY 15, 1988
TIME Is1
0 0'3
3
b
T
T I M E (SI
0
3
Flgue 2. Absorbance signals for a concentrate spiked with 90.9 MIL, diluted four times with HNO, 2% (v/v) (-) and for a standard AI solution with similar concentration (- -): (a) with ammonium nitrate added: (b) without ammonium nitrate added.
25
-
50
100
pg A I / L
Figure 4. Precision at different AI concentrations for peak height (0) and peak area (0)measurements.
3
Table I. Recovery Data for Dialysis Concentrates Spiked with Aluminum
lo
20 p g i L AI
60
LO
*
sample
present
1
8
2
6
AI, a / L added 19.6 47.6 90.9 19.6 47.6 90.9
found 29 54 94 24 57 94
70 recovery 105 97 95 94 106 97 99
&
* 5%"
Mean f standard deviation.
0,200.
-a I
a.
~ c i . 1 0 0-
c
/I/
'C
20
pg L A L
60
LO
Figure 3. (a) Comparison of a callbration line in HNO, 0.2% (v/v) (X) and standard addition llnes for a concentrate spiked with AI 47.6 pg/L and diluted with HNO, 2% (v/v): fourfold dllutlon (0)and twofokl diluHon (0). Integrated absorbances used. (b) Comparison of callbration lines prepared in "0, 0.2% (v/v) (X) and In the concentrate diluted four times with HNO, 2% (v/v) (0). Peak height measurements used.
eliminate interferences due to sodium chloride in the analysis of sea water. The addition of the matrix modifier converts the NaCl into NaN03 and "$1 and these are eliminated at temperatures around 400 "C, whereas for the volatilization of NaCl temperatures above 1100 "C are necessary. Evaluation of the Method. T o check the observation that the addition of ammonium nitrate results in a better removal of the matrix effects, standard addition lines were compared with the calibration line. Figure 3a shows that for the fourfold diluted concentrate a slope which is completely comparable with the slope of the calibration line is obtained. The respective linear regression equations are y = 0.041 + 0.0031~ and y = -0.001 + 0.0031~. It also becomes evident that a fourfold dilution of the concentrate is necessary to completely eliminate the matrix effects since for a twofold dilution the linear regression equation is y = 0.057 0.0019~which has
+
an obvious lower slope than the calibration line. That only with peak area measurements direct determination against a calibration line becomes possible follows from Figure 3b. The peak height measurements are compared for a calibration line in H N 0 3 0.2% and a calibration line prepared in the concentrate diluted four times with H N 0 3 2%. The linear regression equations are, respectively, y = -0.001 + 0 . 0 3 7 ~and y = 0.14 + 0.0752 indicating the strong positive effect of the concentrate on the peak height. The different acid concentrations in both calibration lines was not found to have an influence. As for peak area measurements, standards prepared in H N 0 3 1.5% to match the acid concentration of the samples, gave similar peak height absorbances as did those in HNOB0.2%. The different behavior of the peak area and peak height measurements is explained by the different peak shape for standards and concentrate samples as previously illustrated in Figure 2. Therefore standardization of the analysis against a working curve is only possible if the integrated absorbance is used. Moreover precision was always found to be better with peak area measurements. The precision for peak area and peak height measurements a t different A1 concentrations in dialysate concentrates is shown in Figure 4. Each point is based on six measurements. The precision is better than 10% in the concentration range 20-100 pg/L. Recovery data for two dialysate concentrates spiked with A1 and diluted four times with HN03 2% are given in Table I. The composition of sample 1 is as given previously. The amount of NaC1, NaAc and MgC&in sample 2 is comparable with those of sample 1but the concentration of KCl and CaC1, is smaller (respectively 2.4 and 8.2 g/L). A mean recovery of 99 f 5% is obtained.
303
Anal. Chem. 1900, 60,383-384
In six different dialysis concentrate solutions that we analyzed the A1 concentration ranged between 6 and 22 Mg/L. These are acceptable levels since before use in the dialysis unit a dilution with water is performed. Provided that water with a low A1 concentration is used, hemodialysis solutions with low A1 levels can be prepared. Registry No. Al, 7429-90-5; ",NOB, 7697-37-2.
(5) D'Haese, P. C.; Van de Vljver, F. L.; de Wolff, F. A.; De Broe, M. E. C l h . Chem. (Wlnston-Salem, N.C.)1985, 31, 24-29. (6) Verbeelen, D.; Smeyers-Verbeke, J.; Sennesael, J.; Massart, D. L. Lancet 1803, i , 1168-1169. (7) Mllllner, D. S.; Nebeker, H. G.; Ott, s. M.; Andress, D. L.; Sherrard, D. J.; Alfrey, A. C.; Slatopolsky, E. A.; Coburn, J. W. Ann. Intern. M e d . 1904, 101 775-780. (8) Parkinson. I . S.; Channon, S. M.; Ward, M. K.; Kerr, D. N. S. Trace Hem. Med. 1984, 1 , 139-141. (9) Halls, D. J.; Fell, G. S. Ana/yst((London) 1985, 710, 243-246. (10) Allah, P.; Mauras, Y.; Der Katchadourian, F. Anal. Chem. 1984, 56, 1196-1 198. (11) Edlger, R. D.; Peterson, G.; Kerber, J. D. At. Absorpt. News/. 1974, 13, 61-64.
6484-52-2; HN03,
LITERATURE CITED
.
Smeyers-Verbeke, J.; Verbeelen, D.; Massart, D. L. Clln Chim. Acta 1880. 108. 67-73. Gardher, P. E.; Ottaway, J. M.; Fell, G. S.; Halls, D. J. Anal. Chim. Acta 1981. 128. 57-66.
RECEIVED for review July 7, 1987.
Accepted September 30,
Potentiometric Determination of Halogen Content in Organic Compounds Using Dispersed Sodium Reduction Margaret L. Ware, Mark D. Argentine, and Gary W. Rice* Department of Chemistry, College of William and Mury, Williamsburg, Virginia 23185 We recently became interested in assessing the percent C1 in commercial polychlorinated biphenyls (Aroclors) via classical methods for comparison to spectroscopic techniques being developed in our laboratory for the same purpose (1). A review of the literature revealed a number of classical methods (2), of which most involved tedious and time-consuming combustions, lengthy refluxing, or sodium fusion processes. The specialized equipment and/or glassware required for many of these procedures was not available in our laboratory. A procedure first developed by Stepanow utilized nascent hydrogen, generated from Na in ethanol, to quantitatively displace the halides from organic compounds (3). Several modifications of this procedure have appeared over the years (4-7).A procedure which we thought might be applicable used a dispersed sodium reagent for complete conversion of the organohalogens to free halides (8). The method involved generating sodium alkoxide from a small amount of alcohol added to an inert solvent (e.g., benzene) containing the dispersed Na and organohalogen compound. After a 5-min reaction period, excess Na was reacted with additional alcohol, the solution acidified with nitric acid, and the halide titrated potentiometrically by using silver indicator and glass reference electrodes. The method was reported as being rapid, accurate, and reproducible for a number of aliphatic and aromatic halides. We have been unsuccessful in duplicating these procedures for a number of similar halogenated compounds. The following discrepancies or observations were noted: (1) the aqueous layer (ca. 30 mL) created by the addition of 10% HN03 to neutralize and subsequently acidify the organic layer was insufficient for placement of standard electrodes, even in a tall form beaker; (2) reactions in benzene resulted in a blackish residue at the organiclaqueous interface, which upon stirring coated the electrodes to produce erratic voltage readings; (3) an alternative single phase procedure (concentrated H N 0 3 for acidification) resulted in small amounts of water (where halides would preferentially solvate) clinging to the beaker surface, even with vigorous stirring; (4) results obtained by following the exact procedure were totally unsatisfactory, with poor reproducibility as well as inaccurate 0003-2700/88/0360-0383$0 1S O / O
Table I. Halogen Content Determined for Chloro and Bromo Organic Compounds 70 halide
compound
theory
exptln
70error
1-chloropentane trichloroethylene
33.26
1,2,4,5-tetrachlorobenzen.e
65.69 28.69
33.16 (0.16) 81.11 (0.60) 66.49 (0.50) 28.39 (0.57)
31.22
31.30 (0.12)
27.79 37.91 54.97 54.99 48.41 74.01 67.74 38.59
26.39 (0.26) 38.05 (0.23) 55.98 (0.50) 55.46 (0.32) 48.13 (0.30) 74.05 (0.36)
-0.30 +0.19 +1.22 -1.05 +0.26 -5.04 +0.37 +1.84 +0.85 -0.57 +0.05
46.45
47.25 (0.41)
9,lO-dichloroanthracene 2-chloropyridine p-chloroaniline 2-chloroacetamide 1,3-dichloro-2-propanol dichloroacetic acid 1-bromohexane 1,4-dibromobutane 1,4-dibromobenzene 1-bromonaphthalene p-bromoaniline
80.95
67.23 (0.67)
-0.75
39.59 (0.26)
+2.59 +1.72
Mean from three determinations (& average deviation). values for halogen percentages. This study incorporates several significant modifications of the aforementioned procedure. The halogen content for a number of compounds and complex mixtures has been successfully determined by using the techniques to be described.
EXPERIMENTAL SECTION Reagents and Compounds. HPLC grade methanol (American Scientific) and a 40% (w/v) Na dispersion in light oil (Aldrich no. 21712-3), with anhydrous diethyl ether (Mallinckrodt) as the solvent, were used for the halogen displacement reactions. Silver nitrate solutions (0.15 M) were standardized with 99.999% NaCl (Aldrich). Compounds and Aroclor (polychlorobiphenyl(PCB)) samples (AlltechAssociates) tested for halogen content were used without further purification. Apparatus. A syringe pump (Sage Instruments, Model 341A) was calibrated to give a fixed flow rate of AgN03 from 10- or 30-mL syringes (typically on the order of 1.5 mL/min). The potential generated from standard Ag wire and saturated calomel electrodes was recorded on a strip charge recorder (Fisher Series 0 1988 American Chemical Society
