and Figure 2 present a capillary ethanol concentrationtime profile measured following the administration of 15 ml of 95% ethanol to a single human subject. Choice of Reagents. Sodium fluoride was added to the internal standard solution to inhibit ethanol metabolism (18, 1 9 ) as well as ethanol production (20) by the growth of micro-organisms, Inclusion of the anticoagulant, sodium heparin, in the internal standard solution eliminated the need for heparinized capillary tubes. Sodium nitrite (100 mM) effectively halted the oxidation of ethanol and, therefore, prevented the subsequent increase in acetaldehyde observed a t elevated temperatures (19). Blood samples turned brown-green to black in color when mixed with the internal standard solution containing sodium nitrite; an indication that the oxyhemoglobin was, indeed, converted to methemoglobin (19).
SUMMARY The head-space analysis method described is capable of a routine precision of 4.6% (average) over a 400-fold range of concentration. A sensitivity of 3 kg ethanol/ml is easily reached, and even lower sensitivity is feasible with increased head-space sample size. Since endogenous ethanol levels average 1.2-1.5 wg/ml (21, 22), the lower sensitivity reported is sufficient for blood level studies. Equilibration of the blood ethanol samples a t an elevated temperature resulted in an ethanol enriched head-space gas (14, 23). A change of equilibration temperature from 30 to 60 "C increased the apparent ethanol partition coefficient (air/ blood) an estimated 4-fold. The use of a constant temperature water-jacketed, gas-tight syringe prevented condensation (24) and reduced the scatter previously reported (13) a t elevated temperatures. Storage time had little effect on response at the preferred storage temperature (-17 "C). Appropriate standard ethanol solutions should be used whenever samples are to be analyzed. The assay described should also be suitable for deter-
mining the concentration of ethanol in biological fluids other than capillary blood, such as in urine, serum, or plasma.
LITERATURE CITED J. G. Wagner and J. A. Patel, Res. Commun. Chem. Pathol. Pharmacol., 4, 61 (1972). R. B. Forney. Abstracts of Symposia and Contributed Papers Presented to APhA Academy of Pharmaceutical Sciences at the Meeting of the 118th Annual Meeting of the American Pharmaceutical Association, San Francisco, Calif., March 27-April 2, 1971, Vol. 1 , No. 1 , pp 28-29. R. H. Laessig, Anal. Chem., 40, 2205 (1968). R. B. Forney, F. W. Hughes, R. N. Hager, and A. E. Richards, 0. J. Stud. Alcohol, 25, 205 (1964). N. C. Jain and R. H. Cravey, J. Chromatogr. Sci., 10, 257 (1972). N. C. Jain and R. H. Cravey, J. Chromatogr. Sci., 10, 263 (1972). N. C. Jain and R. H. Cravey, J. Chromatogr. Sci., 12, 214 (1974). R. H. Cravey and N. C. Jain, J. Chromatogr. Sci., 12, 209 (1974). J. A. Hancock, F. L. Mill, and J. R . Miles, Ciin. ToxiCol., 4, 217 (1971). N. C.Jain, Ciin. Chem., 17, 82 (1971). M. K. Roach and P. J. Creaven. Clin. Chim. Acta, 21, 275 (1968). V. J. Perez, T. J. Cicero, and E.A. Bahn, Clin. Chem., 17, 307 (1971). B. L. Glendening and R. A. Harvey, J. Forensic Sci., 14, 136 (1969). R. Bassette. S. Ozeris, and C. H. Whitnah, Anal. Chem., 34, 1540
(1962).
S.Ozeris and R. Bassette, Anal. Chem., 35, 1091 (1963). G. Machata, Clin. Chem. News/., 4, No. 2, Winter, 1972. B. E. Coldwell. G. Solomonraj, H. L. Trenholm, and G. S.Wiberg, Ciin. rowicoi., 4, 99 (1971). G. A. Brown, D. Neylan. W. J. Reynolds, and K. W. Smalldon, Anal. Chim. Acta, 88, 271 (1973). K. W. Smalldon and G. A. Brown, Anal. Chim. Acta, 88, 285 (1973). T. U. Marron and J. J. Hilbe, Proc. Iowa Acad. Sci., 47, 225 (1940). F. Lundquist and H. Wolthers. Acta Pharmacoi. Towicoi., 14, 265 (1958). R. D. Hawkins and H. Kaiant, Pharrnacol. Rev., 24, 67 (1972). R. N. Harger, E. B. Raney. E. G. Bridwell, and M. F. Kitchel, J. Biol. Chem., 183, 197 (1950). D. B. Breimer. H. C. J. Ketelaars, and J. M. VanRossum, J. Chromatogr., 88, 55 (1974).
RECEIVEDfor review January 29, 1975. Accepted May 12, 1975. P.K.W. is a Robert Lincoln McNeil Memorial Fellow, American Foundation for Pharmaceutical Education. Work supported by Grant No. 1R01AA00683-01A1 from the National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health.
Determination of Arsenic and Antimony in Environmental Samples Using Gas Chromatography with a Microwave Emission Spectrometric System Yair Talmi and V. E. Norvelll Analytical Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tenn. 37830
The applicability of a gas chromatograph with a microwave emission spectrometric detector (GC-MES) to the determination of As and Sb in environmentally-based samples Is described. The analytical procedure is based on cocrystalllzatlon of As3+ and Sb3+ with thlonalld and reaction of the precipitate with phenylmagneslum bromide (PMB). Following the decomposition of excess PMB, the trlphenyi arsine and stiblne formed are extracted Into ether and separated on a GC column. Atomic emisslon detection of As and Present address, D e p a r t m e n t of Chemistry, U n i v e r s i t y of T e n nessee, Knoxville, Tenn. 37916.
1510
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
Sb is then accomplished by a MES detector, attached to the GC column outlet. Various parameters affecting the dlgestlon, cocrystalllzatlon, precipitate recovery, and phenylation are discussed, along with some Instrumental problems such as GC column deterlorizatlon and capillary contamination. The detection limits for As and Sb are 20 and 50 pg, respectively, and the relative sensltlvitles are 50 and 125 ng/l. for water samples and 30 and 75 nglgram for solid samples. Samples analyzed included biological and plant tissues, coal and fly ash, and fresh and salt water. The relative error ranges from 1.7 to lq?h and the relative standard deviation from 2.6 to 7.1%.
Gas chromatography, with its excellent separation capability, is highly suited to trace element analysis, provided that the trace materials can be transformed into thermally stable and volatile derivatives and that the detection capability is adequate. Various GC analytical methods for the detection of As through the use of its volatile derivatives, have been reported (2-4). However, none of these were used quantitatively. Recently a new GC method has been described (5). I t involves the extraction of arsenic as its diethyldithiocarbamate complex and phenylation of the complex with a Grignard reagent after solvent evaporation. The triphenylarsine formed is then separated on a GC column and quantitatively determined. The analytical procedure used in this study, was a modification of this method and included the following basic steps: sample digestion, cocrystallization of As3+ and Sb3+ with thionalid (6, 7), filtration and phenylation of the precipitate, and finally GC-separation of the triphenylarsine (Ph3As) and triphenylstibine (Ph3Sb) formed, and their quantitative detection by the microwave emission spectrometric detector (8-21). The present study describes the simultaneous determination of trace-level quantities of As and S b in various environmental samples.
EXPERIMENTAL Apparatus. The GC-MES system, utilized in this study, has been previously described (11, 12). The detection procedure for As and S b involved the GC elution of their corresponding phenylated derivatives (PhsAs and PhsSb) into a low-power microwave, argon plasma and the monitoring of the atomic emission intensity of As and S b a t their corresponding 228.8 nm and 259.8 nm spectral lines. The GC-MES operating conditions are given in Table I. Reagents. Commercially available reagents were used without further purification. PhsAs (Ph=C&), PhsSb, PhsBi and PhnSe, (purchased from Alfa Inorganics) were dissolved in benzene to prepare the necessary GC standards. Standard aqueous solutions of arsenic and antimony were prepared by dissolving metallic arsenic and antimony of AR purity in HzS04 and properly diluting with doubly distilled water. Thionalid (purchased from K&K Laboratories) was dissolved in acetone to prepare a 2% w/v solution. The 2.9M phenylmagnesium bromide (PMB) ether stock solution was purchased from Alfa Inorganics. A 5% aqueous NaCl stock solution was prepared for use in the ionic strength study. T o reduce the blank level, the water and the NaCl solution were cocrystallized with thionalid using the same procedure utilized in the analysis. Radioactive Trace Recovery Studies. The recoveries of As, Sb, Bi, and Se on the filters, after cocrystallization with thionalid, were determined by the use of radioisotopes; 74As,lz5Sb,206Bi,and 75Se,respectively. The solutions necessary for the recovery studies, 0.002-50 pph, were prepared by a proper addition of aliquots of the radioisotopes to their corresponding carriers. Following cocrystallization, both the precipitate and the supernatant solution were counted (well-type NaI scintillation counter) and their y-activity was compared to that of radioisotope controls of equal volume. Digestion, Precipitation, and Filtration of Environmental Samples. Solid samples of coal, fly ash, orchard leaves, and bovine liver, weighing from 0.2-1.0 gram were digested by a HN03-HC104 wet ashing procedure (111, and then accurately diluted to 100 ml. Aliquots (10-50 ml) of these solutions were added to 250-ml beakers and diluted with water to give 100-ml solutions which were also made 1.25% NaCl and 0.1-0.4N HC104. Water samples were acidified with either HC104 or HzS04 (0.2N).With a few modifications, to be discussed later, the thionalid cocrystallization procedure was adopted from Portmann and Riley ( 7 ) . After adding 1 ml of 5% ascorbic acid to the acidified solution, it was heated to boiling (hot-plate) for 3 min to reduce arsenic to its trivalent positive state. The solution was then removed from the hot plate and allowed to cool to room temperature. An additional 2 ml of ascorbic acid and 2 ml of 2% w/v thionalid acetone solution were then added while stirring with a magnetic stirrer. After stirring for 5 min to coagulate the precipitate, the solution was allowed to stand for 10 min. Then the beaker, partially covered with a watchglass was heated to boiling (gently) for 30 min until all acetone was removed. After the solution had been cooled for 30 min (5-10 min with stirring) in an ice bath, it was ready for filtration. Prior to fil-
Table I. GC-MES Operating Conditions Parameter
GC column length a GC column packing
3 ft
Quartz capillary, i .d . , 0 . d. (mm)h Carrier gas C a r r i e r gas flow rate, ml/min Column temperature, "C Injector temperature, "C Capillary heater temperature "C Microwave generator output, watt Monochromator setting Slit height, mm Slit width, pm Wavelength, nm, arsenic, antimony Photomultiplier tube Photomultiplier voltage, V Optics' Lens focal length, mm Lens diameter, mm
0.5, 6.5
4% FFAP on 80/100
mesh Gas Chrom Q argon 100-120 220-240 245-260 3 50
30
8 30 228.8, 259.8 1P28 650
100
50
Pyrex column dimensions, i.d. 3.5 mm; 0.d. 6.5 m m . Quartz capillary length, 16 in. Plasma image was focused on the entrance slit. a
tration of the solution, a homogenous layer of coarse, pure, recrystallized thionalid was collected on the filter to prevent contact between the fine precipitate (sample) and the filter. The layer was formed by passing 30 ml of aqueous thionalid (stock) suspension through the filter. (This stock aqueous suspension was prepared by adding 100 ml of 2% w/v acetone thionalid solution to 1 liter of 1.25% NaCl in deionized water, followed by stirring for 5 min, heating to boiling, and cooling in an ice bath). After partially drying the layer (air suction), the sample was rapidly filtered under vacuum (water aspirator). After filtration was completed, the precipitate was allowed to dry (air suction) for an additional 10 min. I t was then transferred with the filter into a 25-ml bottle and dried in a n oven a t 70 "C for 60 min. When dried, the precipitate (and the thionalid layer) was quantitatively scraped off the filter with a spatula and collected in the bottle. Phenylation of Sample. After the precipitate had been dried a t 70 "C for an additional 10-15 min to ensure the removal of all water, 3 ml of benzene or ether and 1 ml of 1.5M ether PMB solution was added. The bottle was capped (polyethylene-lined caps) and shaken for 50 min. When phenylation had been completed, the excess PMB was decomposed by reaction with 0.5 ml water, and 2 ml of 2% solution of thioglycolic acid in ether were added to prevent oxidation of PhsAs and PhsSb. An aqueous solution of thioglycolic acid could have been added instead. In some phenylationstudy experiments, where large quantities of PMB had been used, cooling in an ice bath was necessary before water was added to minimize ether losses caused by the exothermic reaction. After shaking for 5 min, the sample was centrifuged and aliquots of the organic layer were taken for direct injection into the GC column.
DISCUSSION Analytical Procedures. Chronologically, the analytical procedure consists of the following basic steps: sample pretreatment, cocrystallization and filtration, quantitative removal of the precipitate from the filter, phenylation, decomposition of excess PMB, extraction, and, finally, detection of the phenyl derivatives by the GC-MES. Sample Pretreatment. The HNOS-HC104 wet-ashing procedure was used for solid samples throughout the entire study. Recovery studies, performed with 74As and 125Sb, showed no losses of inorganic As or S b due to either volatility or absorption on the vessel walls. However, the wet-ashing procedure was proved inappropriate for the determinaANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
1511
Table 11. Losses of Antimony Due to Adsorption on Surface of Inorganic Residue Percent recovery lz5Sb (by counting)
Aqueous Samplea
phase
1. Coal 2 . Fly ash 3 . Orchard leaves 4. Coal 5. Fly ash 6 , Orchard leaves
Solid residue
Sb (as PhgSb)
2 99 2 4 95 1 98 3 91 100 90 98 91 99 ... 89 a Samples 1, 2. 3: positive pentavalent lz5Sb and S b were added before wet ashing. Samples 4, 5 , 6: positive pentavalent lz5Sb and Sb were added after wet ashing and separation of the solution from the solid residue.
...
...
Excess amounts of thionalid, serving as both the complexing and crystallizing agent, form water insoluble complexes with submicrogram quantities of As, Sb, and Bi, and bring about their quantitative separation from large volumes of aqueous solutions (less than 10 ng/l.) by cocrystallization. Quantitative separation of arsenic requires that it be reduced to its trivalent positive state. This was achieved by reduction with ascorbic acid at near boiling point temperatures. The pH of the solution must be kept a t or below 1. Both HzS04 and HClO4 can be used for that purpose, although HClOd tends to favor the formation of a finer precipitate. In an early state of this study, it became apparent that the solution has to reach a certain ionic strength in order to achieve a quantitative cocrystallization of As, Sb, and Bi, Table 111. This was accomplished by the addition of alkaline salts, e.g., NaCl, KC1. Maintaining proper ionic strength and cooling of the solution (ice bath) shortened the cocrystallization aging process from the previously rec-
Table 111.Effect of NaCl Concentration and Foreign Ions on the Efficiency of Cocrystallization AS3+,
Volume of
ng/ml
sample, m l
0.1
100 100
0.1 0.1 0.1 0.1
Foreign ions, fig/ml
laC1, 4
0.05 0.20 0.50 0.50 0.50
1uo 100 100
Thionahd,
’
0.04 0.04 0.04 0.04 0.04
HC104, N
1192-
0.2 0.2 0.2 0.2 0.2
... ... ... 1000 ...
Table IV. Effect of Thionalid on the Separation of Precipitate from Filter ’4 recovery of
Thionalid As(V), n g / m l
cocryrtallized
74‘4s
added, mg
40 40
0.4 1.2 0.4 1.2 0.4
45 .O 51.0 100 88 .O 100 86.5 40 + layerb 99.0 1.2 40 + layer 98.5 a Percent recovery was determined after separation of the precipitate from the filter. A layer of Thionalid, 40 mg, was deposited on the filter prior to filtration of the sample.
tion of S b in coal and fly ash since the Sb was absorbed on the undigested inorganic residue, Table 11. Unlike that of selenium ( 1 1 ) , adsorption of Sb was not eliminated upon increasing the volume of the digestive acid mixture. Consequently, the determination of Sb in such samples requires more rigorous digestion methods that will ensure the solubilization of the inorganic residue, e.g., salt fusion. H F acid digestion, however, is not recommended because of losses caused by volatility of AsF3 and SbF3. Cocrystallization a n d Filtration. The major modification in the present analytical procedure over that previously described ( 5 ) , is the use of the “thionalid (thioglycolic6-amino-naphtalide) cocrystallization”, rather than the “diethyldithiocarbamate-extraction” procedure to separate the analyte from the matrix. Use of the “extraction” procedure necessitates the rather complex and lengthy evaporation of the extracting solvent, CHC13, in order to avoid the rapid consumption of the Grignard reagent (13):
- c1
CHC1,
1
:C I
c1
PhMgBr
c1
I
Ph-C-MgBr I
2PhMEBr
C1
Ph,CMgBr 1512
*
H+ -+
Ph,CH
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
? recovery
Fe3i
CU2-
... ... ...
...
1000 e . .
...
e . .
10 100
on filter
60 .O 82 .O 99.5 98.5 0.5
ommended over-night (7) to 30 min. Equally important is the concentration of thionalid in the solution which should be kept above 0.04% (w/v). The interference of some characteristic matrix ions, including: Na, K, Ca, Mg, and A1 was negligible even at the 1000 p/ml level, although the effect of Cu was significant a t much lower concentrations, Table 111. The quantitative cocrystallization of S b and Bi was accomplished under the same conditions as that of As, and therefore all three ions could be determined simultaneously. On the other hand, the recovery of Se(1V) was lower than 85% even at much higher concentrations of thionalid and NaC1. Quantitative recovery was achieved, however, when Hg2+ (above 10 wg/ ml) was added to the solution as a carrier. To minimize NaCl impurity blank, the 5% stock solution of NaCl was purified by the same thionalid cocrystallization procedure used for the samples. Next, the precipitate collection efficiencies of various filters, including Nucleopore and Millipore (0.14- and 0.8-pm pore diameter) and Whatman No. 541 filter paper were tested. The Nucleopore and Millipore filters were found capable of quantitatively collecting the precipitate even at filtration rates above 50 ml/min, whereas the paper filter gave a maximum recovery of only 8085% a t much reduced filtration rates. Even though phenylation could be performed without the removal of the filter, accuracy and precision were significantly improved when the precipitate alone was treated. Unfortunately, strong adhesion of the fine precipitate crystals to the filter prevented complete separation of the two. An increase in the thionalid concentration substantially reduced the adhesion, but easy and quantitative separation of the precipitate was accomplished only when a layer of thionalid (about 50 mg) was deposited prior to the filtration of the sample, Table IV. Phenylation. As previously reported ( 5 ) ,the total conversion of arsenic-diethyldithiocarbamate to Ph3As, using PMB as the phenylation agent, could have been accomplished within 60 min. In this study, however, total phenylation of the cocrystallized arsenic and antimony-thionald, was not achieved, although the yield was adequately repro-
ducible; 88 f 4% and 92 f 4% for Ph3As and PhsSb, respectively. Figure 1 demonstrates the dependence of phenylation yield on time and PMB concentration. T o confirm the assumption that the incomplete recovery of arsenic, as Ph3As, is caused by incomplete phenylation rather than by arsenic losses or inefficient extraction of PhsAs, the following experiments were performed: 1) An As-free water sample (blank) was cocrystallized with thionald. After filtration, a 0.2-hg of Ph3As was spiked into the precipitate which was then reacted with PMB and extracted with ether. All the spiked PhsAs was recovered in the ether, thus indicating that neither the thionalid matrix nor the P M B reagent interferes in the extraction procedure or prevents it from reaching completion. 2) A sample of 75Aswas analyzed by the present analytical procedure. The gamma activity of the radioisotope was monitored in the precipitate, filter, and ether and aqueous layers after phenylation, and in the PhMg(0H)Br residue formed by neutralization of excess PMB with water. 75As was not detected in any of these fractions, except in the ether layer where it was fully recovered. At the same time, the recovery of PhsAs in the ether layer, as determined by the GC-MES, was only 86 f 6% in accordance with previous recovery results. The disagreement between the recoveries of PhsAs and total As, in the ether layer, is probably due to the unaccounted for, unreacted arsenic thionalid complex. 3) Aliquots of an AR-grade Ph3As benzene standard solution were wet-ashed to form As3+ standard solutions, which were then reconverted to Ph3As. The recovery of the phenylated As3+ was determined by comparing the regenerated Ph3As standard with the original Ph3As standard. This procedure eliminated any possible errors due to disagreement between the calibration of As3+ and Ph3As standards. Again the recovery yield of the reconverted Ph3As was 88 4 5%, in agreement with previous results. Other parameters were also studied for their possible effect on the recovery yield; the amount of water or HC1 necessary for the neutralization of excess PMB, the phenylation with and without removal of the filter, and temperature and solvent effects, Table V. Highest recovery yields were obtained with either ether, benzene, or toluene as solvents at room temperature. Recovery yields were significantly lower when using xylene or T H F as solvents or when working with ether at higher temperatures. Once the excess P M B reagent was destroyed, a further addition of H20 or HCl solution had no effect. Instrumental Parameters. Column Deterioration. The continuous injection of unpurified ether extracts caused a
l o ) P H E N Y L A T I O N T I M E , minutes
I
IO I
20
30
I
40
50
60
I
i
201
I I 1 I 0.1 0 2 0.3 0.5 0.5 ( e ) CONCENTRATION OF PHENYLMAGNESIUM BROMIDE, M
Figure 1. Dependence of
phenylation yield on time and PMB concen-
tration slow accumulation of a nonvolatile residue in the upper portion of the GC column (inlet) that eventually reduced the sensitivity and precision of the measurements. This problem was eliminated by replacing the upper (2-inch) portion of the column solid support every 150-250 injections. After every 600-800 injections, a new GC column was used. Capillary Contamination. Volatile organic impurities in the samples are introduced into the capillary with each injection (unnoticed by the highly selective detector), and gradually form a brown deposit which reduces the transparency and thus the intensity of the light emitted through the capillary walls. Usually a few injections of air, with the plasma operating a t atmospheric pressure, or injections of water at 5-10 Torr, eliminated this problem by oxidizing off the deposit. Once a week, however, it was advantageous to remove the capillary and clean it by sequential rinsing with concentrated HF, H20, and acetone, a 5- to 10-min procedure. Following these maintenance procedures, a single capillary was used over a five-month period without significant deterioration in its performance. Measurements. Selectivity and Sensitiuity. Free atoms of arsenic and antimony produced in the plasma are excited and detected by monitoring their emission intensities a t the 228.8-nm and 259.8-nm spectral lines, respectively. The selectivity ratio of the 228.8-nm emission line for arsenic, when Ph3As is compared to PhsSb, is in excess of 20,000. The selectivity ratio of the 259.8-nm line for antimony, when Ph3Sb is compared to Ph$As, is lower, 2000. Thus, since the MES detector is highly selective of As and Sb, sample cleanup procedures were not necessary. Also, a complete separation between the solvent and the analyte chromatographic peaks was not essential. The GC operat-
Table V . Effect of HzO, HC1, Filter Presence, a n d Temperature on Ph3As Recovery &3+,
H20,
Parameter studied
u9
ml
rnl
Filter
H2O HC1 H20 + HC1 HZO H2O Filter presence" Solvent
0.2
0.3
...
... ...
Temperature
0.2
...
0.4 0.4
0.3 1.o
0.4
5 .O 1.o 1.o
(0.2 (0.2 0.4 0.4 0.4 0.4 0.4
1.o 1 .o
1.o
1 .o 1.o
P h d s peak
HC1 (0.1 N),
0.3 0.3
... ... ... ... ... ... ... ... ...
...
*..
...
Millipore
Nucleopore Nucleopore
...
*..
*..
...
Solvent
Ether Ether Ether Ether Ether Ether Ether THF~
Toluene Xylene Benzene Ether (50 "C)
Phgs,
height, arbi-
ug/ml
wary units
0.05 0.05 0.1 0.1 0.1 0.05 0.05 0.1 0.1 0.1 0.1 0.1
25 26 51 52 52 19 24 18
50 41 51 38
a The effect exerted by the presence of the filter (45 mm diameter) on the phenylation process. * With tetrahydrofuran (THF) as a solvent, an unknown component (non-arsenic) is also eluted which can cause the slow contamination of the capillary.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
1513
RIPHENYLARSINE BENZENE SOLUTION 0.8 ng (0.46p g / m l i
DETERMI NATION OF ARSENIC AS TRIPHENYLARSINE (228.8 nm As Line I
DETERMINATION OF ANTIMONY AS TRiPHENYLSTlEiNE ( 2 5 9 8 nm Sb Line)
1.8 ng 3.32 ,ug/ml)
-
O.Z#'qml TRIPHENYLSTIBINE STANDARD SOLUTION
TRIPHENYLARSINE STANDARD SOLUTION
h
h
@.3pq/rnl
Sb-SPIKED
"15 ORCHARD LEAVES
0.4 ng (o.o8pg/ml)
N0S ORCHARD LEbVES L 1571 1
0.4 ng (0.1 Lg/ml)
-
I
NBS ORCHARD L E A i E S ,#I5711
5b-SPIKED SEA WATER
4
0.2 ng (0.08 ,ug/m'
TIME,
Figure 3. Determination of Sb and leaves
LL TIME,
H
ICOG~NDS As
in sea water and orchard
Spectroscopic Considerations. Microwave ( M W ) Power. The immediate noticeable effect of an increase in the MW power output is generally the elongation of the discharge in both the up- and down-stream directions. Also, the brightness of the discharge and the background emission intensity (and thus the noise) are both enhanced and carbon deposition on the capillary walls becomes more severe. At the same time, the emission intensities of As and S b are practically unaffected by an increase in the power output from 30 to 100 watts. Therefore, throughout the entire study, the power output was kept at the 30-watt level. Different relations between intensity and incident MW power were observed in another study (14),where a similar MES detector was interfaced to an arsine generator. There, the effect of MW power on the detected intensity did indicate a linear increase in the signal with increasing power up to 100 watts and leveling off thereafter. This apparent disagreement is most probably related to the fact that the two MES detectors were operated at different argon flow rates:
2 minutes
Flgure 2. Chromatograms of triphenylarsine benzene solution
ing conditions were, therefore, selected to maximize the sensitivity and minimize the analysis time. The detection limit, defined as the amount of the element that produces a signal twice the magnitude obtained from the thionalid blank, was approximately 20 pg for As and 50 pg for Sb. Since a substantial preconcentration gain (100) can be provided by the cocrystallization procedure, water samples could be determined with relative sensitivities of 50 ngll. for As and 125 ngll. for Sb. With solid samples, however, the relative sensitivity was only 30 ng/gram for As and 75 nglgram for Sb.
Table VI, Determination of Arsenic in Environmentally Based Samples Sample
Orchard Leaves, NBSNo. 1571 Bovine Liver, NBSNo. 1571
Average sample
No. of samples
size
analyzed
Reported or hewn Concn of As found, ppm
0.25 g
16
11.5
5
0.3
0.30 g
2
0.1
*
0.01
0.15 g
8
0.20 g
9
6.1 i 0.3
4 4
0.0023 0.00780
concn of As, ppm
11
No. 1 No. 2 As5 spiked
150 ml 150 ml
60
i
NBS (uncertified)
0.00012 0.00030
*
6
NBS
5.9 f 0.6
NBS
61
3
*
0.0025 0.0075
+-
sea water No. 1
(I
200 ml 3 No. 2 200 ml 4 Spiked As5- was added to the sample prior to digestion.
1514
0.00053 0.00160
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
+
*
NBS
2
0.59
Fly Ash," NBS No. 1633 Coal,' No. 1632 As5'- spiked drinking water
i
0.000030 0.000085
Independent source or analytical method used
0.00050 0 .OO 150
i
Addition of known amount of As5'
Table VII. Determination of Antimony in Environmentally Based Samples No. of
Sample
Orchard Leaves, NBS No. 1571 Sbs'-spiked drinking water No. 11 No. 2 \
Sb"-spiked sea water No. 11 No. 2 ) a
Average sample
samples
size
analyzed
Reported or h o r n Concn oi Sb found, ppm
0.2 g
5
3.2
100 ml 100 ml
3
0.0038 0.0072
4
i 0.18
i
Independent source or analytical method used
concn of Sb, ppm
3.14
0.0002
f
NAA'
0.13
0.0035 0.0075
i 0.004
Addition of known amount of sb5+ 250 ml 250 ml
3 3
0.00018 i 0.00001 0.00042
f
0.00020 0.00040
0.00003
SeeRef. (16).
the GC-MES at 120 ml/min and the arsine generator-MES a t 600 ml/min. As the flow rate increases, more MW energy is required to sustain the argon plasma and to ensure its efficiency as a spectrometric source, Le., to optimize the dissociation and excitation parameters. Thus, with the low flow rate plasma, this optimal state (emission signal leveling off point) is reached at a lower MW power. Variations i n Emission Intensity along t h e Discharge. Study of the variation in the spectral emission intensities of As, Sb, and Bi, as a function of the position along the discharge plume has shown a definite maximum a t the extreme upstream region, 1-5 mm, of the discharge. In fact, monitoring of S b and Bi a t downstream regions was impossible due to erratic fluctuations in their emission intensities, especially a t the chromatographic retention time (peak maximum) when the concentration of the analyte in the discharge is maximal. This interference increased with the metallic nature of the hetero atoms, from As to S b to Bi. When a large sample of PhsAs (or S b and Bi) was injected, a metallic deposit started to form just above the analytically optimal upstream zone. All these observations suggest that the fragmentation of the phenyl compounds and the subsequent excitation of the As, Sb, and Bi free atoms formed is rapidly accomplished a t the first contact with the plasma. The signal fluctuations and the loss in sensitivity observed in downstream regions is caused by interaction of the metal atoms with the relatively cold quartz walls, leading to their eventual removal from the discharge. This behavior was previously observed with Cu and A1 chelates (15),and might set a limit to the applicability of the GC-MES technique to trace metal analysis. Analytical Results. Analytical Working Curves. The emission intensity response, measured as peak height was linearly proportional to the analyte concentration in the 0.01-1 wg/ml range utilized in this study. Linearity and injection reproducibility are shown in Figure 2. As observed previously ( I I ) , the linearity of the working curves was strongly dependent on the chromatographic parameters and only slightly on the operating conditions of the plasma. As discussed above, the recovery of As and S b were incomplete, but adequately reproducible, so that concentrations could be determined by using the proper recovery correction factors. T o reconfirm the validity of these factors, a known sample of As and S b was analyzed along with each batch of samples. The results obtained in this manner were in good agreement with those determined by the standard addition method, although they were less accurate. Analysis of Environmental Samples. To demonstrate the analytical feasibility of the method, various samples were studied including: bovine liver, orchard leaves, coal and fly ash-all NBS standard reference materials and fresh and salt water samples, Figures 2 and 3. Coal and fly
i
COLUMN TEMPERATURE
240T INJECTION TEMPERATURE
260'C 40
60
80
IW
250'C
260°C
260-C
270'C
120 IS0 TIME FROM INJECTION l s e c I
Figure 4. Reconstructed Sb, and Bi
160
180
chromatogram of phenylated Se,
Te, As,
ash samples were not analyzed for their S b content for reasons explained above. As shown in Tables VI and VII, the relative error for arsenic and antimony varied from 1.7 to 10% with an average of 5.2%. It should be noted, however, that the error values assume the validity of the certified values supplied. Regarding precision, the average relative standard deviation was 5.2% and ranged from 2.6 to 7.1%. Scope. Preliminary studies with Se, Te, and Bi suggest that it should be possible to determine all three simultaneously with As and Sb. Figure 4 shows a reconstructed chromatogram of all five phenylated elements. It is comprised of five individual chromatograms, separately obtained for each element at five different wavelengths. T o actually achieve a simultaneous determination of all five elements, a computer-programmed scanning monochromator or a multichannel spectroscopic detector would be desirable in order to monitor the emission intensity a t the different wavelengths. In addition, a temperature programming of the GC column will also be necessary. LITERATURE CITED (1) J. Tadmor, J. Gas Chromatogr., 2,385 (1964). (2) D. Vrant-Piscou, J. Kontoyannakos, and 0 . Parissakis, J. Chromatogr. Sci., 0, 1499 (1971).
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
1515
(3) R. S. Juvett, Jr., and R. L. Fisher, Anal. Chem., 37, 1752 (1965). (4) W. C. Butts and W. T. Rainey. Jr., Anal. Chem., 43, 1538 (1971). (5) G. Schwedt and H. A. Russel, Chromatographia, 5, 242 (1972). (6) M. G. Lai and H. V. Weiss, Anal. Chem., 34, 1012 (1962). (7) J. E. Portmann and J. P. Riley, Anal. Chim. Acta. 31, 509 (1964). (8)A. J. McCormack, S.C. Tong, and W. D. Cooke, Anal. Chem., 37, 1470 (1965). (9) C. A. Bache and D. J. Lisk, Anal. Chem., 38, 1757 (1966). (10) H. Kawaguchi, T. Sakamoto, and A. Mizuike. Talanta, 20, 321 (1973). (11) Y. Talmi and A. W. Andren, Anal. Chem., 46, 2122 (1974). (12) Y. Talmi, Anal. Chim. Acta, 74, 107 (1975). (13) S. L. Eck, Kalarnazoo College, Kalarnazoo, Mich., 1974, private cornmunication. (14) F. E. Lichte and R. K., Skogerboe, Anal. Chem., 44, 1480 (1972).
(15) R. M. Dagnal, T. S.West, and P. Whitehead, Analyst, (London),08, 647 (1973). (16) A. R. Byrne, Anal. Chim. Acta, 59, 91 (1972).
RECEIVEDfor review February 27,1975. Accepted April 22, 1975. Research supported by the National Science Foundation-RANN Program under NSF Interagency Agreement No. 389 with the U.S.Energy Research and Development Administration. Oak Ridge National Laboratory is operated for the US.Energy Research and Development Administration by the Union Carbide Corporation.
Determination of Amitriptyline at Nanogram Levels in Serum by Electron Capture Gas-Liquid Chromatography Jack E. Wallace,’ Horace E. Hamilton,’ Linda K. Goggin,’ and Kenneth Blum2 University of Texas Health Science Center at San Antonio, San Antonio, Texas 78284
An analytical method for the quantitative determlnation of amitriptyllne/nortriptyline in small amounts of biologlc fluids at the nanogram level Is described. The procedure involves the oxidation of the drugs to a polyaromatic carbonyl derlvatlve, anthraquinone, that has the intrinsic capability to capture electrons with an efflciency comparable to that of polyhalogenated compounds. The electrophilic property of the product provides the basis for the extreme sensitivity of the procedure. Oxidation is accomplished by employment of a solution of cerlc sulfate-sulfuric acid that selectively oxidlres only the ethylene group of the lO,ll-dlhydro-5-dlbenzo[ a,d]cycloheptene moiety of the molecule. The product is measured by electron capture gas-liquid chromatography. Quantitation of the method is extensively enhanced by utilization of ethylanthraquinone as the internal standard.
Amitriptyline and its monomethylamino analog, nortriptyline, are among the therapeutic agents most commonly administered for the amelioration of depression. It is established that the tricyclic antidepressants demonstrate marked pharmacokinetic heterogeneity which promotes differences in plasma levels for subjects receiving equivalent dosage regimens (1-6). Identical dosages may result in subtherapeutic levels in certain subjects and toxic levels in others (1, 4 , 5 ) . Most investigators, therefore, consider the measurement of circulating blood levels to be of significant clinical importance during tricyclic antidepressant therapy (5, 6). It has been observed ( 5 ) that combined plasma amitriptyline and nortriptyline (a principal metabolite of amitriptyline and a commercially available therapeutic) levels exhibit greater correlation with the clinical management of depression than do plasma levels of either amitriptyline or nortriptyline alone. Earlier spectrophotometric methods (7-9) for the determination of the tricyclic antidepressants lacked a certain level of sensitivity for application to the determination of the drugs in serum or plasma. The authors of this report (10) recently described a spectrophotometric procedure for the determination of amitriptyline and nortriptyline that D e p a r t m e n t of Pathology. D e p a r t m e n t of Pharmacology.
1516
ANALYTICAL CHEMISTRY, VOL. 47, NO. 9, AUGUST 1975
provided sufficient sensitivity for the measurement of blood levels if both amitriptyline and nortriptyline were analyzed simultaneously a t high therapeutic and toxic levels. Sensitive methods for the specific determination of nortriptyline are available utilizing isotopic labeling with subsequent scintillation spectrometry ( 1 1 ) or polyfluoroacetylation with subsequent quantitation by electron capture gas-liquid chromatography (12). However, these latter techniques are not applicable to the analysis of amitriptyline which has a tertiary amine group at the functional nitrogen. A number of gas-liquid chromatographic (GLC) methods for the determination of amitriptyline and nortriptyline have been reported in recent years. Hucker and Miller (13) described a technique which, by application of the Hoffman reaction, converts both of these tricyclics to a common derivative prior to GLC determination. Their report did not contain any biologic data, and the data presented were for a concentration range greater than that encountered in plasma specimens. Braithwaite and Whatley (14) reported a GLC procedure for amitriptyline in urine, and Norheim (15) described a GLC method for the determination of amitriptyline and nortriptyline in biologic specimens. The latter procedure provided a limit of sensitivity of 1 pg/ml for a 5-ml specimen. GLC methods with a 20 ng/ml limit of detection for a 5-ml plasma specimen have recently been described for the simultaneous and separate determination of amitriptyline and nortriptyline (16, 1 7 ) . These procedures require a unique chromatographic packing and a special treatment of the packed column (16) or are somewhat tedious and time consuming, requiring additional solvent wash, back-extraction, and evaporative techniques not required in the method proposed in this report ( 1 7 ) . Currently available GLC procedures for the tricyclic antidepressants, with the exception of Walle and Ehrsson’s procedure (12) for the analysis of nortriptyline, utilize flame ionization detectors. This report depicts a rapid electron capture gas-liquid chromatographic procedure capable of detecting 1 ng/ml or reliably quantitating 5 ng/ml of amitriptyline/nortriptyline in 0.5-ml plasma or serum specimen within one hour of the time the sample is received at the laboratory. The procedure is dependent upon the oxidation of the parent compounds to anthraquinone which is subsequently chromatographed.
