Silver halides can be used to advantage over Ag silicates because of their better thermal and UV stability. Unlike the silver silicates, the silver halides are presumably only adsorbed on the Si02 surface and are blocking part of the active adsorption centers. As a result, one observes decreased and more even adsorptive forces (lower retention) and a smaller capacity. The supports behave similarly to many commercially available chemically bonded stationary phases for HPLC. The silver halide supports showed an improved selectivity for the compounds tested (aza arenes and picolines). A direct relationship between retention of acridine and the concentration and type of silver halide has been observed, which follows theoretical expectations. The complexing influence is also evidenced by the dominance of steric factors over basicity for the separation of aza arenes (12) and picolines. The halide supports are probably too weak for selective separations of unsaturated compounds. This may be due to the limited loadability of Si02 with AgX. Using a support as proposed by Houx et al. ( 1 4 ) based on fixation of Ag+ on ion-exchanger materials would likely eliminate this problem.
LITERATURE CITED (1)R. S.Mulliken. J. Phys. Chem., 56, 801 (1952). (2)0. K. Guha and J. Janak, J. Chromatogr. (Rev.), 68,325 (1972). (3)V. Prey, A. Berger, and H. Berbalk, Z.Anal. Chem., 185, 113 (1962). (4)C. Vidai-Madjar and G. Guiochon, J. Chromatogr. Sci., 9,664 (1971). (5) K. Yasuda, J. Chromatogr., 74,142 (1972). (6)K. Shirnornura and H. F. Walton, Sep. Sci., 3, 493 (1968). (7)R . L. McCreery and D. T. Sawyer, J. Chromatogr. Sci., 8, 122 (1970). (8)W. Schurig, J. L. Bear, and A. Zlatkis, Chromatographla, 5, 301 (1972). (9) F. Mikes, W. Schurig, and E. Gil-Av. J. Chromatogr., 83, 91 (1973). (IO)D. Kunzru and R. W. Frei, J. Chrornatogr. Sci., 12, 191 (1974). (11) M. Kraitr, R. Korners, and F. Cuta, Anal. Chem., 46,974 (1974). (12)R. Vivilecchia. M. Thiebaud, and R . W. Frei, J. Chromatogr. Sci., 10, 411 (1972). (13)R . W. Frei, K. Beall, and R . M. Cassidy, Mikrochim. Acta, 859 (1974). (14)N. W. H. Houx, S.Voerrnan, and W. M. F. Jongen, J. Chromatogr., 96, 25 (1974). (15)R. M. Cassidy, D. S. Le Gay, and R. W. Frei, Anal. Chem., 46, 340 (1974). (16)“Grnelin’s Handbuch der Anorg. Chernie”, Band 3,“Das Silber”. (17)L. R . Snyder, “Principles of Adsorption Chromatography”, Marcel Dekker, New York, N.Y., 1968.
RECEIVEDfor review August 7, 1975. Accepted September 19, 1975. The “Jubilaumsfond” of the Austrian National Bank is to be thanked for an instrument grant in support of this research.
Mobile Phase Effects on Atomic Absorption Detectors for High Speed Liquid Chromatography D. R. Jones IV, H. C. Tung, and S. E. Manahan” Department of Chemistry, 123 Chemistry Building, University of Missouri, Columbia, Mo. 6520 1
Atomic absorption spectrometry (AAS) Is a sensitive, selective detector for organometallic compounds and metal chelates separated by high speed liquid chromatography (HSLC). This comblnatlon requires special consideration of flow rates and mobile phase effects on AAS signals obtained when the aspirator Is fed at a controlled rate from the HSLC column. The nature and flow rate of solvent affect the reducing power and/or temperature of the flame. The AAS signal does not decrease linearly with decreased flow rate and is, in fact, enhanced proportionately to the rate of Introduction of analyte metal at lower flow rates. This is explained on the bask of the free atom fraction In the flame. These effects are discussed for Cr, Cu, Co, and Zn in water mobile phase and Cr, Cu, Co, NI, and Fe in 2-propano1, 2butanol, CHC13, and ethyl acetate.
In a preliminary paper ( I ) , the combination of a liquid chromatography system with atomic absorption detection was reported. Subsequent study has indicated the wide usefulness of the detector system for applications of liquid chromatography involving metals. The high sensitivity and specificity of the detector are its greatest strengths, and will be discussed in some detail in this paper. Considerations of the application of this detector system must be divided into two categories based on the constitution of the mobile phase. Organic and aqueous mobile phases will be considered separately.
EXPERIMENTAL Apparatus. Three different liquid chromatographs were used for this work. The first was a “homemade” helium driven displacement pump, constructed along the lines of Felton (2). (Use of a high pressure regulator allowed helium pressures up t o 2000 psi t o drive the mobile phase.) This liquid chromatograph is, of course, a constant pressure device. The second pump used was a Spectra-Physics, model 740B (Spectra-Physics, Inc., Auto Lab Division, Santa Clara, Calif.). The pump is a dual piston reciprocating model, capable of pressures to 7000 psi and flows to 20 ml per minute. T h e pump is of constant flow design, with a flow-feedback transducer to maintain constant flow. Each piston displaced 500 p l per stroke. The third pump used in the study was a Waters, model 6000A (Waters Associates, Inc., Milford, Mass.). This is also a dual piston reciprocating pump, capable of pressures to 6000 psi and flow rates to 10 ml/min. I t is also of the constant flow design, utilizing a pulse damper system and a nonsymmetric cam. Each piston displaces 100 gl per stroke. The atomic absorption spectrophotometer (AAS) used in this study was a Perkin-Elmer model 403. A 4-inch single slot burner, and a nitrous oxide burner head were used for the varying phases of the study. Flames used were either air-acetylene or acetylenenitrous oxide. Flame characteristics will be discussed in a later section. A Spectra-Physics 10 gl, loop-type sample injection system was employed to introduce samples onto the column. A variety of columns were used, and they will also be discussed in a later section.
DISCUSSION Atomic absorption spectrophotometry possesses several unique advantages as a detector for high speed liquid chroANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
7
Table 11. Burner Response for Aqueous Mobile Phases0
Table I. Nebulization Efficiencies for Atomic Absorption Detector0 Vol. i n t o flame per t i m e , ml/min
Chromatograph
flow,
Metal
ml/min
Peak heightb,c
Percent Decrease
Dd
Cr Cr cu cu
0.8 0.4 0.8 0.4 0.8 0.4 1.o 0.5
66.5 53.5 184 126 82.5 75 181.6 140.9
19.5
0.064
31.5
0.98
Flow rate, ml/min
Amount discarded, mlb
Percent nebulized
Ethanol
2.0 1.0 2.0
1.4 0 1.7
86 100 83
Chloroform
2.0
co co
2.0 1.0
Zn Zn
Mobile phase
Methanol
Benzene
1.7 1.0 1.6
1.0 1 .0
0 100 1.0 Water 2.0 7.4 26 0.5 1.0 6.8 32 0.3 0.5 6.2 38 0.2 0.3 2.6 74 0.2 Flame conditions: air, 33.5 psi, 10.4 l./min; C,H,, 8.8 psi, 1 0 . 3 l./min. Flame conditions kept the same for all aspirated fluids. b 10 ml of fluid was aspirated into t h e burner, and the fluid from the drain tube was collected and Q
9.1 22.4
0.052 0.45
Distilled water mobile phase, air-acetylene flame. b Average of five readings, units arbitrary. C Metals levels for the various elements were adjusted to give a conveniently measured peak height. Absolute metal levels are not given, since the table is intended for ordinal comparisons only. d Free atom fraction, from Ref. 3. Q
measured.
matography (HSLC). Chief among these is the extreme selectivity of the detector. In complex samples containing many different chemical species, the ability of AAS to detect only a pre-chosen metal gives the chromatographer a powerful analytical tool. When using conventional LC detectors such as ultraviolet absorption or refractive index, numerous trace organic impurities will give a detector response. These detectors may not selectively discriminate against unwanted peaks; they may yield a high background signal or even fail to detect the desired peak when it is very small. While limited in application to those species containing (or capable of chelating) metals, AAS as a detector for LC provides the analyst with a sensitive, selective detector, and eliminates many problems of detector interferences. Sample pretreatment to remove interfering organic species prior to running the chromatograph becomes unnecessary. Applications of this detector to separations involving metallopharmaceuticals and organometallics in petroleum and coal liquids are two examples of uses where the specificity of the detector system will aid the analyst in achieving the desired separation. Analyses involving metal-containing enzymes and other biological systems containing metals are examples of further applications of this metalspecific detector. Peaks which elute simultaneously with the desired species, and would interfere in conventional detectors will not be “seen” by the AAS detector. Rigorous requirements for solvent purity required by UV and RI detectors, with their concomitant prices, are not necessary when using the AAS detector. The solvents must be free of particulate matter, as in all LC applications, but bulk grade solvents will generally meet these requirements. The major thrust of this work was an examination of the phenomenon taking place in the burner flame when atomic absorption is used as a detector for metals in HSLC. T o eliminate any effects from the column while examining the behavior of metals in the flame, columns were used in this work which provided no retention of the species investigated. While contrary to conventional chromatographic techniques, this method allowed the examination of burner behavior, without imposing any column variables on the study. Therefore, any phenomenon noted was flame-connected rather than being based on any column parameters. With these conditions in mind, the data were examined on the basis of composition of mobile phase. Aqueous Mobile Phases. When aspirating water solu8
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
tions into the AAS burner, the response of the detector was dependent on several variables. The most important of these was flow rate of the chromatograph into the AA burner. For maximum sensitivity, the Perkin-Elmer aspirator used in this work required a flow rate of between 2 and 3 ml/min. This was the amount of liquid the aspirator would draw through a tube from an open beaker of water. The distinction must be drawn between the amount of liquid the burner will draw, and the smaller amount normally supplied to it by the liquid chromatograph in this study. Hence, the term, “starving the burner” is appropriate. Nebulization efficiencies (measured as the amount of liquid nebulized and swept into the flame per unit volume of liquid aspirated) a t these flow rates were on the order of 25%. (See Table I.) As the flow rate from the chromatograph to the aspirator decreased, the nebulization efficiency increased. At 1 ml/min, the efficiency was 3296, and a t 0.5 ml/min it was 38%. In each of the latter cases, the aspirator was “starved” for liquid, that is, it was receiving less liquid than it would by its own suction in the case of unrestricted fluid flow. In order to test the sensitivity of the AAS detector, a 10-pl sample of a metal solution was injected onto the liquid-chromatographic column. The flow rate of the LC pump was varied, and the AAS response monitored on a recorder. Data for several metals in water mobile phase are presented in Table 11. The peak heights and areas are in arbitrary units intended for comparison of the data from one metal only. Comparisons between different metals are tenuous a t best, since the sensitivity of AAS of course varies from metal to metal. I t can be noted that there is a general decrease in peak height as the flow rate is decreased, but the relationship is not linear. This fact seems to stem from several factors, among them nebulization efficiency. As the flow rate decreases, less metal (in the ionic form) is delivered to the nebulizer per unit time, but a larger portion of the metal ions arriving are nebulized. The increase in nebulization efficiency does not account, however, for all of the effect noted, so one must also consider the temperature quenching effect of the water vapor on the flame, and the decrease of the free atom fraction corresponding to the lower flame temperatures. This effect should be related to the ease with which the particular metal is atomized, since metals with lower free atom fractions should show a greater temperature dependence. This can be demonstrated in the case of copper and chromium in Table 11. When copper samples were injected a t flow rates of 0.8 and 0.4 ml/min, the peak heights were 184 and 126 units, respectively. This is a decrease in peak height of 31.5%. Similarly, chromium
samples injected a t the same flow rates gave peak heights of 66.5 and 53.5, respectively, a decrease of 19.5%. The differences in peak height decrease can, we feel, be credited to the large difference in free atom fractions between the two metals. de Galan and Winefordner's ( 3 ) figures for the free atom fraction of chromium and copper are in air-acetylene flames and are 0.064 and 0.98, respectively. Thus it would seem that minor temperature variations should have a much greater effect on the atom population of chromium than that of copper, a thesis which seems to be borne out by the work done a t this laboratory. The effects noted are also true for cobalt and zinc, where the decrease in peak height with decreasing flow rate is dependent on the free atom fraction in the flame. Thus cobalt, whose free atom fraction is given as 0.052 by the same authors, shows only a 9.1% decrease in peak height, as would be expected. Correspondingly, the results for zinc showed a decrease in peak height as 22.4%. Thus, the relationship between peak height decrease and flow rate decrease is demonstrated to be a function of the (3 (free atom fraction) value for the element being assayed. In an aqueous mobile phase, the flame quenching of higher flow rates and subsequent temperature decrease seems to be the major effect influencing sensitivity of the detector. The consequence of this effect is that sensitivity of the AAS detector for HSLC separations does not decrease as rapidly as one would expect, thus allowing separations a t chromatographically more efficient low flow rates without undue losses in sensitivity. The use of this detector for metals whose free atoms fractions are quite high, such as copper, ( p = 0.98) does pose the limitation of more marked decrease in sensitivity at low flow rates. We then have the interesting phenomenon of getting better detector performance a t low flow rates from metals not normally thought of being among the best for conventional AAS work. Organic Mobile Phases. The effects noted a t the burner-aspirator with aqueous mobile phases described above are even more marked when organic mobile phases are used and aspirated into the AAS burner. The increased sensitivity for atomic asborption analysis of metal chelates in organic mobile phases has been previously noted (4-6). Some of the increased sensitivity is due to the marked increase in nebulization efficiency when organics are aspirated (See Table I). Generally, with solvents less polar than ethanol, the nebulization efficiency was 100% for the flow rates used in this study. This increase in efficiency of aspiration obviously introduces more metal ions into the burner of the AAS, which increase sensitivity; however, the effects noted in this work when organics are used as mobile phases are not adequately explained by increased vaporization alone. For selected elements, there is an increase in sensitivity far above that which would be expected. When aspirating organics, the vaporized mobile phase enters into the flame chemistry as either a quenching agent, if it does not burn, or as a source of fuel if it is combustible. Since, in the case of most combustible organics, the heat of combustion will be less than that if acetylene is the sole fuel, lower flame temperatures result. This tends to lead to less atomization (lower free atom fraction) of metals for those species which are quite temperature sensitive. As the flow rate of the combustible organic solvent is decreased, the flow of acetylene is increased to maintain insofar as possible the same flame characteristics. This results in a hotter flame, which in turn atomizes more of the metal. Balancing this effect is the fact that the amount of metal being pumped into the nebulizer at the lower flow rates is less per unit time. Were the flame conditions not to change, one would expect to see a large decrease in signal response as the flow rate was lowered. The actual response is much
Table 111. Burner Response for Organic Mobile Phases0 Metal
Cu Cu Cu Cu Cr Cr Co Co Co Co
Flow ml/min
0.8
0.4 0.8
Mobile phase
2-Propanol 2-Propanol 2-Butanol 2-Butanol
Peak heightb,"
91.2
Peak height decrease, %
29.8
pe
0.98
64
88.56 28.2 63.5 CHCl, 74.25 -20d 0.064 0.4 CHC1, 89.1 1.0 CHCl, 66.35 16.8 0.5 CHC1, 55.15 0.052 1.0 Ethyl acetate 53.8 16.9 44.7 0.5 Ethyl acetate Ni 1.0 CHC1, 125 25 Ni 0.5 CHC1, 93.75 Fe 1.0 CHC1; 97.5 0.66 Fe 0.5 CHCl, 72.6 25.5 Air-acetylene flame, single slot burner. b Av. of five readings, arbitrary units. C Metals levels for the various elements were adjusted to give a conveniently measured peak height. Absolute metal levels are not given, since the table is intended for ordinal comparisons only. d Negative because lower flow gave higher peak (See text). e Free atom fraction, from Ref. 3. 0.4 0.8
higher than expected, primarily because of this increase in flame temperature as flow rate decreases. If one were to decrease the flow by half, then one half as much metal per unit time would be pumped into the burner, and the signal response should be 50% of the original, higher flow rate. In practice, the lower flow rate yields a decrease of only about 20%, depending on the particular metal used. (See Table 111.) Therefore, there is a large and noticeable enhancement of the signal a t the lower flow rates. It seems the higher flame temperatures are chiefly responsible for this effect, since the nebulization efficiency is 100% for these flammable organic solvents. With nonflammable organics such as chloroform or carbon tetrachloride, the flow-rate effects are similar to those involving water, where quenching of the flame seems to be the most important effect. A notable deviation from the above behavior was found when the metal being assayed was chromium. When aspirating organics, either flammable or nonflammable, and monitoring chromium with the AAS, the behavior of the sensitivity as a function of flow rate was opposite that of other metals. As flow rate was decreased, there was a marked increase in sensitivity (See Table 111).The key to understanding this phenomenon seems to be the rich reducing flame used for AAS of chromium. Increased sensitivity for several metals in fuel-rich flames has been noted (7-9); however, the effect noted here seems to be related to the ratio of fuel (acetylene) to combustible mobile phase. The presence of organics in the fuel-rich flame seems to lessen both the reducing power of the flame and its temperature. The effect of this is to yield a rise in sensitivity as flow decreases from flows of 2 ml/min to 0.2 ml/min. The relative peak heights for these two flow rates, for example, were 65.9 and 106.9, respectively. Therefore, it can be seen that chromium behaves differently than the other metals tested in this study, which may be attributed to the reducing effects of the fuel-rich flame used in AAS for chromium. This effect could not be duplicated for other metals in this study, as they did not seem to be as sensitive to the reducing power of the fuel-rich flame as chromium was. Difficulties in duplicating this enhancement effect were also encountered because of the limited availability of organometallics in the elements requiring a fuel-rich flame (Sn, Ca). The enhancement effect was not noted in the nitrous ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
9
Table IV. Detection Limits for Selected Metals Metal
A q u e o u s Mobile Phase Min. c o n c n Absolute detection (mg/l.)asb limit, n g
Fe Cr
0.72 2.14
7.2 21.4
Zn
0.50
5.0
0.13
1.3
cu
Organic Mobile Phase
Fe 0.79 7.9 Cr 0.67 6.7 co 0.60 26.0 a Based on the concentration of material necessary to produce a signal twice the standard deviation of the random background noise. b Based on 10-pl sample.
oxide-acetylene flame for any of the metals assayed, including chromium. The temperature change in the nitrous oxide-acetylene flame when aspirating different quantities of liquids apparently shows little or no effect on the free atom fraction in the flame. Detection Limits. In order for the AAS detector system to be useful in liquid chromatographic work, the detector must be both selective and sensitive. The selectivity and sensitivity of conventional AAS work has been well noted ( 1 0 ) . This selectivity also applies to AAS when used as an LC detector and, while the range of compounds for which it is suitable is not as wide as other LC detectors, its specificity far surpasses that of other detectors. Difficulties can arise when using the detector for assays of several different metals, since lamp and burner changes are necessitated. Chromatograms of several species containing the same metal are, however, ideally suited for the detection system. The detection limits of this system were measured in the conventional manner, i.e., the concentration of material necessary to produce a signal twice as large as the standard deviation of the random background noise. The detection limits for several of the transition metals in both organic and aqueous phases are presented in Table
IV. These values are applicable under the experimental conditions used and may be improved under more nearly optimum conditions. The conditions of the AAS, such as sensitivity and gain of the electronics, and adjustment of concentration control on the Perkin-Elmer AAS may allow lower detection limits for selected metals. Generally speaking, solutions containing metals in the mg/l. (part per million) range are satisfactory for this type of detection. The use of organic mobile phases requires some modification of the AAS for optimum performance. When aspirating organics, the three-slot burner head gave significantly better results than did the standard single-slot burner. A "hotter" (lower fuel-air ratio) flame could be generated with the three-slot burner, because of its ability to cope with a higher gas flow without excessive instability. The absolute detection limits in Table I11 are based on a 10-pl sample. Use of a larger sample loop would, of course, lower these figures.
LITERATURE CITED (1) D. R. Jones IV and S. E. Manahan, AnalLeff., 8 (a), 569 (1975). (2) H. Felton, J. Chromatogr. Sci, 7 , 13 (1969). (3) L. de Galan and J. D. Winefordner, J. Quant. Spectrosc. Radiat. Transfer, 7, 251 (1967). (4) J. A. Dean, "Flame Photometry", McGraw-Hill, New York, N.Y., 1960. (5)R. Mavrodineanu and H. S.Boiteux, "Flame Spectroscopy", Wiley, New York, N.Y., 1965. (6) S.Gomiscek and M . Span, Anal Chim. Acta, 69, 49 (1957). (7) V . A. Fassel and V. G. Mossotti, Anal. Chem., 35, 252 (1963). (8) J. D. Winefordner and C. Veillon, Anal. Chem., 36, 943 (1964). (9) Walter Slavin and D. C. Monning, Anal. Chem., 35, 253 (1963). (10) G. W. Ewing, "Instrumental Methods of Chemical Analysis", 4th ed., McGraw-Hill, New York, N.Y., 1975.
RECEIVEDfor review August 25, 1975. Accepted October 13, 1975. This research was supported by National Science Foundation Grant No. MPS75-03330 and United States Department of the Interior Office of Water Research and Technology Matching Grant B-095-MO. The atomic absorption spectrophotometer was purchased in part by funds provided by a National Science Foundation Research Instrumentation Grant.
Determination of 1,4-Benzodiazepines and -diazepin-2-ones in Blood by Electron-Capture Gas-Liquid Chromatography J. Arthur
F. de Silva,"
lhor Bekersky, Carl V. Puglisi, Marvin A. Brooks, and Robert E. Weinfeld
Department of Biochemistry and Drug Metabolism, Hoffmann-La Roche Inc., Nutley, N.J. 07 7 10
Electron-capture gas-liquid chromatography (EC-GLC) was applied to the determination of a number of 1,4-benzodiazepine drugs classified into groups based on their extractability into benzene-methylene chloride (9O:lO) at specified pH values. Derivatization by means of either methylation at position Nq-H or silylation at the 3-hydroxy position was employed where necessary for the analysis of certain 1,4-benzodlazepin-2-ones in order to reduce either adsorption or thermal rearrangement losses and to yield symmetrical peaks of high sensitivity. The compounds were determined as either the Intact 1,4-benzodiazepine or -diazepin-2-ones in blood. These methods were used to measure blood concentratlons following single oral therapeutic doses of diazepam, bromazepam, clonarepam, and flunltrazepam with 10
ANALYTICAL CHEMISTRY, VOL. 48, NO. 1, JANUARY 1976
a sensitivity limit of the order of 1.0 to 10.0 ng of compound/ mi of blood. The assay is simple, dlrect, and readily automatable for large scale clinical analysis.
The 1,4-benzodiazepine class of compounds has yielded several important drugs ( I ) which are widely used in clinical practice (2).The pharmacology of these compounds has been extensively studied and their anti-anxiolytic, sedative, muscle relaxant, and anticonvulsant activity is well documented ( I , 3). Of these compounds, chlordiazepoxide and medazepam are the 1,4-benzodiazepine type, whereas diazepam, flurazepam, nitrazepam, clonazepam, flunitrazepam, bromazepam, temazepam, oxazepam, and lorazepam are of the 1,4-benzodiazepin-2-onetype. The chemical
