(9)A. J. Barnes,Rev. Anal. Cbem., 1, 193 (1972).
(IO) L. Andrews, Ann. Rev. Phys. Cbem., 22, 109 (1971).
(11) J. S. Ogden and J. J. Turner, Cbem. Br., 7 , 186 (1971). (12) B. Meyer, Science, 168, 783 (1970). (13) A. J. Barnes and H. E. Hallam, Q. Rev., Cbem. SOC., 23, 392 (1969). (14) G:Mamantov, E. J. Vasini, M. C. Moulton, D. G. Vickroy, and T. Maekawa, J. Chem. fbys., 54, 3419 (1971). (15) A. Snelson, J. Phys. Cbem., 73, 1919 (1969). (16) M. M. Rochkind, Spectr‘ocbim.Acta, Part A, 27, 547 (1971). (17) M. M. Rochkind, Anal. Cbem.. 39,567 (1967). (18) H. E. Hallam and G. F. Scrimshaw, in Ref. 3, pp 29-31. (19) P. R. Griffiths, “ChemicalInfrared Fourier Transform Spectroscopy”,John Wiley, New York, N.Y., 1975. (20) H. W. Schnopper and R . I. Thompson, Methods Exp. Pbys., 12A, 491 (1974). Rev., 4, (21) J. Chamberlain, G. W. Chantry, and N. W. B. Stone, Cbern. SOC. 569 (1974). (22) J. L. Koenig, Appl. Spectrosc., 29, 293 (1975). (23) J. Cuthbert, J. Phys. E, 7 , 328 (1974). (24) P. R . Griffiths, C. T. Foskett, and R. Curbelo, Appl. Spectrosc. Rev., 6,31 (1972). (25) National Academy of Sciences, “Particulate Polycyclic Organic Matter”, Washington, D.C., 1972. (26) H.-L. Boiteau, M. Robin, and S. Gelot, J. Eur. Toxicol., 293 (1972).
(27) J. L. Monkman, L. Dubois, and C. J. Baker, Pure Appl. Cbem., 24, 731 (1971). (28) C. Pupp, R. C. Lao, J. J. Murray, and R. F. Pottie, Atmos. Environ., 8, 915 (1974). (29) M. J. Linevsky, J. Cbem. Pbys., 34, 587 (1961). (30) L. Brewer, G. D. Brabson, and B. Meyer, J. Cbem. Pbys., 42, 1385 (1965). (31) H. E. Hallam and G. F. Scrimshaw, in Ref. 3, p 30. (32) E. L. Wehry, G. Mamantov. R . R . Kemmerer, H. 0. Brotherton, and R. C.
Stroupe, in “PolynuclearAromatic Hydrocarbons: Chemistry,Metabolism, and Carcinogenesis”,R. I. Freudenthal and P. W. Jones, Ed., Raven Press, New York, N.Y., 1976, p 299. (33) R. N . Perutz and J. J. Turner, J. Cbem. SOC.,Faraday Trans. 2, 69, 452
(34) (35)
(1973). H. E. Hallam, in Ref. 3, p 68. H. 0. Brotherton, G. Mamantov, and E. L. Wehry, unpublished work
RECEIVEDfor review August 23,1976. Accepted October 22, 1976. This work is supported by the Electric Power Research Institute, Contract 741011-RP-332-1. The purchase of the FTS-20 spectrometer was partially supported by the National Science Foundation Research Instrument Grant GP-41711.
Evaluation of Pyrolytic-Graphite-CoatedTubes for Graphite Furnace Atomic Absorption Spectrometry R. E. Sturgeon and C. L. Chakrabarti’ Metal Ions Group, Department of Chemistry, Carleton University, Ottawa, Ontario K 1s 5B6,Canada
A comparison of the analytical characteristics of pyrolyticcoated-graphlte tubes and uncoated tubes for use in the Heated Graphlte Atomizer 2100 has been made. Uncoated tubes compare favorably with respect to sensitivity, detectlon limit, and precision (for both the peak and lntegratlon modes of absorbance measurement) for Cd, Zn, Cu, AI, and Sn. Slgnificant improvementsin performance from the coated tubes were obtained for Mo and V. It is suggested that this Is due to the lower poroslty of the pyrolytic-graphite coatlng, resulting in reduced soaking of these elements Into the pores of the surface at high temperature.
L’vov (1)has advocated the use of pyrolytic graphite for the construction of electrothermal atomizers for many years. This material, obtained from the high temperature pyrolysis of hydrocarbons, has the properties of low permeability to gases, low porosity, high purity, a higher sublimation point (3970 K) than standard graphite, a high resistance to oxidation, and a high thermal conductivity. Atomizers constructed of pyrolytic graphite thus offer the advantages of improved vapor confinement, less “soaking” of the analyte into the graphite surface pores, increased lifetime, and uniform heating. As a result of these factors, an improvement in sensitivity, detection limit, and precision in graphite furnace atomic absorption may be expected when pyrolytic graphite is substituted for conventional graphite. Atomizers made of standard graphite and coated with a layer of pyrolytic graphite can also be expected to provide such improvements (1-3). Despite the number of advantages offered by pyrolytic graphite, many instrument manufacturers have failed to make use of this material for the construction of atomizers. Recently Perkin-Elmer Corporation described and evaluated a technique for coating HGA2100 tubes with pyrolytic graphite by an in situ pyrolysis of a mixture of 10% methane-90% nitrogen ( 4 ) . This study was undertaken to evaluate the effects of a coating of pyrolytic graphite over spectrographic graphite on 90
ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
the atomization and analytical characteristics of the absorbance signals produced in a Perkin-Elmer Heated Graphite Atomizer 2100.
EXPERIMENTAL Apparatus. Pyrolytic-graphite-coated tubes (0.001-in. thick coating of PT-101pyrolytic graphite) for the Perkin-Elmer HGA-2100 were supplied by Ultra Carbon Corporation (Bay City, Mich.). Con-
ventional tubes, obtained from the Perkin-Elmer Corporation were subsequently coated with pyrolytic graphite by Ultra Carbon Corporation. In addition,a number of tubes which had been coated with pyrolytic graphite according to the procedure described by Manning and Ediger ( 4 ) ,were supplied by the Perkin-Elmer Corporation. All other instrumentation (integration control unit and detector-recorder system) has been described elsewhere (5, 6 ) . All electron micrographs were taken with a JEOL model JSM-U3 Scanning Electron Microscope. Reagents. Stock solutions of Zn, Cu, AI and Sn, 1000 pg/ml, were prepared from the metals, Cd from the carbonate, and V and Mo from the pentoxide and trioxide, respectively. The above metals or compounds were dissolved in pure acids or bases, where required, and diluted with ultrapure water obtained from a Milli-Q water system (Millipore Corporation).All test solutions were prepared immediately prior to use. Gases. High purity (99.95%)argon and nitrogen gases were used to sheath the graphite tube. Procedure. Sample volumes of 5 pl were introduced into the HGA-2100 furnace with an Eppendorf syringe fitted with disposable plastic tips. The furnace was operated in the internal purge gas interrupt mode, with Ar as the sheath gas. The effect on the signal of substitution of Nz for Ar was examined for the atomization of Al. Lamp currents, wavelengths, and spectral bandwidths used in this study have been presented in a previous paper ( 5 ) .
RESULTS AND DISCUSSION The useful lifetime of a graphite tube is determined principally by the atomization conditions under which it is operated (i.e., steady-state temperature of the furnace attained as well as its duration) and the analyte element. The analyte element determines the wavelength at which spectroscopic measurements are made and, hence, the amount of scattering
Table I. Effect of Atomization Temperature on Peak and Integrated Absorbance
Element Cd
I.o
I
20
30
40
Zn
10
TIME, s
Figure 1. Temperature-time characteristics of an uncoated tube (A) and a coated tube (B)
of the source radiation by carbon particles ejected from the incandescent graphite surface. In addition to the wavelength, the physicochemical properties of the analyte element may also affect the useful lifetime of a graphite tube. This is particularly evident with the relatively involatile elements, some of which may also form refractory carbides. As the tube ages and becomes more porous, the extent of carbide formation may increase. At such a point, the tube may suffer from memory effects, irreproducible atomization as well as light scattering, Whereas the tube may become useless for further analytical determinations of such an element, it may perform quite satisfactorily with other, more volatile elements. Consequently, it is difficult to present a definite figure regarding the lifetime of a particular tube. The useful lifetime of the graphite tubes is greatly extended if the tube is coated with pyrolytic graphite. More than 100 atomization cycles a t a maximum temperature of -3000 K maintained for a period of 10 s (during the atomization of Mo and V) with an Ultra Carbon coated tube, failed to produce as much visible wear (powdery surfaces) or light scattering as 20 such cycles with an uncoated tube. To obtain accurate comparisons between the performances of the coated and uncoated tubes, their electrical characteristics were considered. With the contact tension between the furnace tube and the cones set to an arbitrary value, the measured resistance of the furnace assembly was found to be 29 mQ for the pyrolytic-graphite-coated tubes processed by Ultra Carbon Corporation, 22 mQ for the uncoated tubes, and 20 mQ for those pyrolytic-graphite-coated tubes prepared in situ by the method prescribed by Manning and Ediger ( 4 ) .All subsequent measurements taken in this study were made with a furnace assembly resistance of 22 mQ (contact tension was increased when the Ultra Carbon coated tubes were used, in order to reduce the furnace resistance from 29 mil to the above value; similarly, the tension was reduced for the Perkin-Elmer coated tubes). Under such conditions, the power consumed by the atomizer and, hence, the rate of heating of the furnace a t an arbitrary current setting on the control unit, will be determined solely by the physical and electrical characteristics of the tubes. In both cases, the pyrolytic-graphite-coated tubes showed slower rates of rise of temperature than the uncoated tubes. Figure 1 shows the temperature-time profiles of the atomizer surface (obtained with an automatic optical pyrometer) for an uncoated (A) and an Ultra Carbon Corporation coated tube (B) a t an atomization setting of 2700 OC (reset for each tube). The heating characteristics of the coated tubes supplied by Perkin-Elmer were intermediate between curves A and B. The power delivered t o the atomizer is related to the resistance of the heating element, R , and the current drawn, I , by
cu A1 Sn Mo
v
Temperature setting, “C”
Absorbance Integrated, absorbance X Peak seconds 0.253 0.271 0.266 0.235 0.192 0.344 0.327 0.292 0.170 0.262 0.225 0.092 0.260 0.175 0.238 0.182 0.051 0.320 0.125 0.250 0.145 0.043
2700 2100 1700 1000 800 2700 2100 1700 1400 2700 2500 2100 2700 2500 2700 2500 2100 2700 2500 2700 2500 2300
0.118 0.216 0.322 0.525 0.484 0.188 0.267 0.289 0.290 0.173 0.174 0.128 0.175 0.107 0.431 0.370 0.164 0.462 0.446 0.315 0.281 0.145
a The temperature setting has been shown in the centigrade scale in Agreement with that on the meter scale of the atomization control unit.
P
= 12R
Although the power consumed is a direct measure of the rate of heating of the atomizer, additional factors, such as the mass of the tube, must also be taken into account in order to explain the results shown in Figure 1. The resistivity of pyrolytic graphite is much greater than that of regular, high-purity graphite ( 7 , p 98). The decreased resistance of the PerkinElmer coated tubes may be rationalized by considering the low resistance graphite tube and its interior “jacket” of high resistance pyrolytic graphite to be a parallel electrical circuit of two resistors, the net resistance being lower than that of an uncoated tube. The tube from Ultra Carbon Corporation has a layer of high resistance pyrolytic graphite deposited over both the length of the tube and the ends which make electrical contact with the cones. The anisotropy ratio of specific resistance (c-axisla-axis) is about lo4 in pyrolytic graphite obtained by cracking CH4 on a heated carbon surface at 2400 K ( 7 ,p 98). Because of the orientation of the planes of graphite in the pyrolytic coating, parallel to the coated surface, the power consumption is reduced since this requires that the current flow be perpendicular to these planes a t the surface in contact with the cones. Because of its lower resistance, the Perkin-Elmer coated tube will heat up more slowly than an uncoated tube. This has been experimentally confirmed. In contrast to the predictions of Equation 1,however, the high resistance coated tube from Ultra Carbon Corporation heats up at the slowest rate. The reason for this anomalous behavior lies with the mass of this graphite tube. The masses of the coated and uncoated Perkin-Elmer tubes are roughly the same, whereas that of the Ultra Carbon tube is almost 25% greater. The result is a larger effective heat capacity tending to decrease the rate of heating of these tubes. Since the rate of heating of the graphite tube directly influences the analyte peak absorbance ( I , 5, 6,8-IO), the decreased rate of rise of temperature given by pyrolytic-graphite-coated tubes is a ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
91
Figure 2. Oscilloscopic trace showing absorbance by 1.5 X aluminum in a pyrolytic-graphite-coated tube
g
(A) Ar sheath gas; (B) N2 sheath gas. Vertical scale: absorbance, O.l/scale unit. Horizontal scale: sweep speed, 500 ms/scale unit
Figure 3. Oscilloscope trace showing absorbance by 1.5 X aluminum (uncoated tube)
g
(A) Ar sheath gas; (B) N2 sheath gas. Vertical scale: absorbance, O.l/scale unit. Horizontal scale: sweep speed, 500 ms/scale unit
drawback in analytical use. Data concerning the performance of pyrolytic-graphite-coated tubes reported throughout the remainder of this paper refer exclusively to those obtained with the Ultra Carbon coated tubes. The performance characteristics obtained for the Perkin-Elmer coated tubes are in agreement with those reported by Manning and Ediger ( 4 ) . Characterization of Absorption Pulses. The atomization “temperature” (as indicated by the meter on the HGA-2100 control unit) and time were individually adjusted to maximize both the integrated and peak absorbance for each element. The effect of the atomization temperature on the peak and the integrated absorbance for elements atomized from the Ultra Carbon coated tubes is presented in Table 1. The behavior of the signals with respect to increasing rates of atomization (higher temperatures) is identical to that which is observed when uncoated graphite tubes are used. These effects, which have been rationalized on the basis of the kinetics of formation and dissipation of analyte atoms within the furnace, have been described in a previous paper ( 5 ) . Aluminum shows a radical change in its atomization behavior in coated tubes. Figures 2 and 3 are absorbance-time traces for the atomization of A1 from both a coated and an uncoated tube. Two peaks are obtained with the coated tube, 92
ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
corresponding to the sequential atomization of two distinct species of significantly different volatility. This behavior was observed for both the chloride and the nitrate forms of the analyte. The substitution of N:! for Ar as a sheath gas greatly reduces the peak absorbances of both peaks and the total integrated absorbance. This effect may be attributed to the formation of involatile aluminum nitride ( I l ) , rather than an increase in the diffusional losses resulting from this substitution of sheath gas. In its outermost layers, graphite can form lamellar compounds with water (7, p 176). Recent work by Frech and Cedergren (12, 13) indicates that hydrogen is formed a t high temperatures (-1000 K) by the reaction of graphite with surface-adsorbed water remaining in the tube after the drying step. The amount of hydrogen produced by a regular graphite tube was found to be considerably larger than that produced by a tube coated with pyrolytic graphite. This is presumably due to differences in the adsorptive capacities of the two surfaces for water and their intrinsic reactivity. In the absence of a hydrogen diffusion flame, two distinct absorption peaks are obtained for A1 in a Varian Techtron Carbon Rod Atomizer (CRA) model 63 (coated tube), whereas, in the presence of a hydrogen diffusion flame, only a single peak occurs ( 5 ) .This observation, along with those made by Frech and Cedergren (12,13) leads to the conclusion that the radical change in the atomization behavior reported here for aluminum in a pyrolytic-graphite-coated tube may be related to the inability of this tube to produce sufficient hydrogen. The larger amount of hydrogen generated by an uncoated tube plays a role in the chemical elimination of one of the aluminum peaks. Atomization of Mo and V is greatly improved when pyrolytic-graphite-coated tubes are used. Despite the slower rates of rise of temperature available with these tubes, both elements could be completely atomized a t temperature settings as low as 2500 and 2300 OC, respectively (Table I). This is a significant improvement over uncoated tubes, in which temperature settings below 2700 “C result in incomplete atomization of these elements and, hence, “memory effects”. Optimum atomization conditions and the corresponding pulse characterization times for the atomization of elements from the coated tubes are presented in Table 11. The characterization times correspond to the times elapsed from the point a t which the atomization cycle begins to the points in time corresponding to the appearance, the peak, and the completion of the absorbance signal ( 5 ) .The display time ( 5 ) , which is the total duration of the pulse, is also presented in Table 11. The major difference between the coated and the uncoated tubes is the decreased display times for Mo and V obtained with the uncoated tubes (cf. 6). This is a significant advantage in analysis as shorter atomization and integration times would be adequate for these elements. This aspect is particularly important when the long-term stability of the source is poor and, in the absence of a double-beam system, baseline drift may become significant, thus limiting the precision and accuracy of the integrated measurements. The melting points, boiling points, and the temperatures of the graphite tube surface a t the characterization times for each element (peak absorbance mode) are presented in Table 111. The appearance temperatures are, within experimental error, equal to those obtained when uncoated tubes are used (cf. 6). Whereas the temperatures corresponding to the absorbance peak and the end of the signal are directly dependent on the kinetics of growth and decay of the atom population (which is determined by the geometry of the atomizer, the permeability of the graphite tube to the atomic vapor, and the rate of heating of the atomizer), the temperature a t the appearance time is relatively insensitive to these factors (11). This is a consequence of the necessity of a minimum atomic
Table 11. Pulse Characterization Times under Optimum Conditions
Element Cd (peak) Cd (int) Zn (peak) Zn (int) CU"
Ala Sna Moa V"
Atomization conditions Temp., Time, "C S 2100 1000 2700 1700 2700 2700 2700 2700 2700
Characterization Times, ms 7 appearance
Tpeak
Tend
Tdisplay
300 800 700 600 850 1750 1100 1800 1850
620 2050 1020 1650 1510 2030 2200 2700 2350
1600 7700 2050 3300
1300 6900 1350 2700 3250 3550 6900 6200 4650
3 9 3 4 5 7
9 9 8
4100
5500 8000 8000 6500
For both the peak and the integration modes.
Table 111. Pulse Characterization Temperaturesa Element Cd Zn cu AI Sn Mo V
Temperature, K
Melting point, K
Boiling point, K
Tappearance
Tpeak
Tend
594 693 1356 933 505 2890 2173
1038 1180 2840 2740 2543 4885 3653
780 1140 1280 2070 1500 2100 2140
1070 1430 1880 2270 2370 2620 2460
1960 2280 2920 3000 3000 3000 3000
Corresponding to the surface temperatures of the furnace at an atomization setting of 2700 "C.
population accumulating within the graphite tube before a signal can be detected, this population being governed by the vapor pressure-temperature characteristics of the element or the strength of the chemical bond broken during the atomization of the analyte (11).A more comprehensive discussion of the significance of the pulse characterization temperature8 may be found in earlier publications of the present authors (5, 6). Effect of S u r f a c e S t r u c t u r e of T u b e on Atomization. The ease of atomization of Mo and V in pyrolytic-graphitecoated tubes may be attributed to their low surface porosity. Porosity is defined as the percentage by volume of the pores in relation to the total volume. The porosity of pyrolytic carbon is zero, as compared to 17% for regular high-purity graphite ( 1 4 ) . Figures 4-7 are electron micrographs of the interior surfaces of unused and used pyrolytic-graphite-coated and uncoated tubes. The surface of the coated tube is much smoother and less porous than that of the uncoated tube. The nodular appearance of the surface is a characteristic growth feature of the pyrolytic coating. This occurs as a result of surface imperfections in the substrate graphite and the possibility that supersaturation levels of carbon vapor cause gas-borne nuclei to form and grow during the coating process (15). The uncoated tubes show a cavernous structure of disoriented microcrystals. After extensive use, little change in the appearance of the surface is evident with the coated tube (compare Figures 4 and 6) whereas there is significant alteration of the surface of the uncoated tube, its crystalline structure developing a loose and extremely porous spongelike appearance (compare Figures 5 and 7). The increased surface porosity of uncoated tubes allows for increased penetration of the analyte into the pores a t high atomization temperatures (1,16-19). Once the analyte has penetrated the graphite surface, atomization becomes more difficult and is accompanied by well-known
"memory effects" (18).As the tube ages, these problems become progressively more acute (compare Figures 5 and 7). Pyrolytic graphite has the advantage of low surface porosity with the result that the sample does not penetrate into the walls of the tube and atomization of relatively involatile elements is greatly facilitated. Not only may lower rates of rise of temperature be employed in such cases, but the time taken for complete atomization (the duration of the absorbance signal) is also significantly reduced (Tables I and I1 and (5)). When Mo is atomized from a pyrolytic-graphite-coated tube, an atomization energy of 160 kcal mol-', corresponding to the heat of atomization of the metal (157.5 kcal mol-'), is obtained from an E , plot (11).This is in agreement with the value of 165 kcal mol-' obtained when Mo is atomized from an uncoated tube (11).It follows, therefore, that the mechanism of atomization of Mo (11) remains unchanged when pyrolytic-graphite-coated tubes are used. It is reasonable to conclude, therefore, that the radical change in the atomization behavior of Mo (and V) from coated tubes is due only to the decreased porosity of the pyrolytic graphite surface, as discussed earlier. Figure 8 is an absorbance-time trace of V a t an atomize setting of 2700 "C. Curve A was obtained using a pyrolyticgraphite-coated tube supplied by Ultra Carbon Corporation, curve B a regular graphite tube (obtained from another supplier) which was subsequently coated with pyrolytic graphite by Ultra Carbon Corporation. It is evident that the tube which gave curve A had superior performance characteristics. Not only was the atomization time of the analyte greatly decreased with this tube, but the sensitivity was higher. The reason for this probably lies in the difference between the two coatings. Residence Times. The residence times, (defined as the time taken for the absorbance signal to decay from its maxiANALYTICAL CHEMISTRY, VOL. 49,
NO. 1,
JANUARY 1977
* 93
Figure 4. Electron micrograph of the surface of a new PYrOlYtiCgraphite-coated tube, X500
Figure 6. Electron micrograph of the surface of a used pyrolytic150 atomization cycles at 2700 "C), graphite-coated tube (after
=
X500
Figure 5. Electron micrograph of the surface of a new uncoated tube,
X500 94
ANALYTICAL CHEMISTRY, VOL.
49, NO. 1, JANUARY 1977
Figure 7. Electron micrograph of the surface of a used uncoated tube (after 1130 atomizations at 2700 "C),X500.
Table IV. Atomization Times, 71,and Residence Times, 7 2 Coated tube Element 71, ms 72,ms Cd (peak) 320 580 Cd (int) 2100 1500 Zn (peak) 320 290 Zn (int) 1050 600 Cua 660 440 A1a 280 300 1100 Sn a 1000 900 Mo 850 1580 V" 500 0
ms 320 1600 280 1450 930 760 820
71/72
71,
0.55 1.40 1.1001 230 1.75 1.50 0.93 1.10 1.05 0.32
Uncoated tube 72, ms 310 1000 0.82 1100 1140
1180 700
8300 3600
1100 1100
71/72
1.03 1.60 1.32 0.82 0.65 1.17 0.13 0.30
For both the peak mode and the integration mode.
Table V. Absolute Sensitivitya Mode of Measurement Element Cd Zn
cu
AI Sn Mo V
Coated tube Peak (oscilloscope) Peak (recorder) 5.4 x 10-13 2.2 x 10-13 8.7 X 1.1 x 10-11
7.9 x 10-11 1.1x 10-1' 4.6 X 10-l'
5.6 x 10-13 2.9 x 10-13 8.8 x 1.7 X lo-" 8.5 x 1.1x 10-11 5.0 X 10-l'
Integration 3.3 x 10-13 4.1 x 10-13 1.1 x 10-11
2.7 X 4.7 x 10-11 6.3 X 2.7 X 10-l'
Peak (oscilloscope) 3.5 x 2.6 x 2.1 x 1.7 x 2.9 X 8.8 x 4.0 X
10-13 10-11 lo-"
10-l1
Uncoated tube Peak (recorder) 3.0 X 2.1 x 1.9 x 3.3 x 8.8 x 4.0 x
Integration 4.2 X lo-',' 3.7 x 10-13
4.0 X 10-11
10-11 lo-" 10-l'
1.1 x lo-" 1.1 x 10-1'
1.9 x 10-11 9.1 x 10-12 5.2 x 10-11
Defined as the slope of the analytical working curve; the weight of element in grams which gives a peak absorbance of 0.0044 or an integrated absorbance of 0.0044 absorbance X second.
mum to l/e of this value ( I , 5 ) ) and atomization times (the time taken for the absorbance signal to rise from the baseline to its maximum value ( 5 , 6 ) )obtained , for the elements in both tubes, are presented in Table IV. In each case, the reported values are those obtained under optimum conditions both for the peak and the integration modes of measurement. Contrary to expectation, the residence times for most elements are significantly shorter in the coated tubes. The decreased permeability of the pyrolytic graphite should improve the vapor containment, leading to residence times longer than those obtained with an uncoated graphite tube. This apparent discrepancy may be resolved by considering the nature of the atomization process in the HGA-2100. Atomization (the production of atomic vapor) of the analyte occurs as the temperature of the furnace rises; in no case can atomization be accomplished under isothermal conditions. In addition, the graphite-tube furnace is a complex kinetic system (6) to which L'vov's concept of the atomization time 71 (being the time taken for the complete "introduction" of the analyte into the analysis volume ( I , 5 ) and, hence, equated to the time taken for the absorbance signal to reach the maximum), cannot be rigorously applied. In the case of less volatile analytes, atomization does, in fact, extend beyond the time required for the absorbance maximum to be attained, the absorbance maximum simply marking the point of balance between the rates of supply and dissipation of analyte vapor within the analytical volume (defined by the geometry of the light beam). The tail of the absorbance pulse, from which 7 2 is evaluated, contains a contribution from the continuing atomization of the analyte, causing the signal to fall from the peak in a less than exponential manner (diffusional loss at constant temperature). In addition, the tail is also influenced by the nonisothermal nature of the diffusion medium. As a result of these two perturbations, the 7 2 values provide little support for a mass-dependent diffusional loss mechanism. Both of the above effects must be considered in order to rationalize the
Figure 8. Oscilloscopic trace showing the absorbance by 5.0 X
g vanadium (A) Coated tube obtained directly from Ultra Carbon Corporation:(B) tube obtained
from another supplier and coated by Ultra Carbon Corporation. Vertical scale: absorbance, O.l/scale unit. Horizontal scale: sweep speed, 500 mslscale unit
behavior of the residence times in the coated and uncoated tubes. I t is difficult to take the rising temperature of the graphite tube into account because of the large uncertainty in the vapor temperatures. Beyond the absorbance maximum, the vapor temperature severely lags behind that of the graphite tube surface, in some cases by as much as 900 K (20). As the influence of the rising temperature on the tail of the absorbance pulse is, to a first approximation, somewhat the same for both tubes (Figure l),the reason for the decreased 7 2 values in the coated tubes probably lies in the difference between the atomization processes in the two tubes. With the exception of Sn, all of the elements (in the peak absorbance mode) have shorter atomization times 7 1 in the coated tubes ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
95
Table VI. Detection Limit, Gram" Coated tube
Uncoated tube
Element
Peak
lntegration
Peak
Cd Zn cu A1 Sn Mo V
1.3 X 2.6 X 1.2 x 10-11 5.7 x 10-11 1.9 x 10-10 6.9 X lo-" 2.9 x 10-10
5.8 x 10-13
3.0 X 2.7 X 1.4 X loW1' 2.6 X lo-" 2.4 X 2.4 X 3.6 X
1.8 X 1.8 X lo-" 6.6 X lo-" 2.0 x 10-10 7.7 x 10-11 2.3 X
Integration 3.7 x 1.9 x 2.0 x 3.1 X 2.0 x 2.1 x 3.4 x
10-12 10-12
10-1' lo-" 10-1' 10-10
10-10
Defined as the weight of the analyte in grams that gives a signal equal to twice the standard deviation of 10 replicate measurements with the same solution containing a very low concentration of the analyte element. (I
(Table IV). This is particularly evident for the more difficult-to-atomize elements. The reduced 71 values, a consequence of the decreased penetration of the analyte into the pores of the graphite surface, directly influence the residence times. Provided atomization can be rapidly completed, the contribution of the atomization time to the tail of the pulse will be negligible, resulting in a wholly diffusion-dependent decay (across a rising temperature profile). If atomization continues beyond the peak absorbance, the decay of the signal will be less than that expected for purely diffusional decay of the maximum population (that at the absorbance peak). The measured 7 2 values will therefore reflect not only the true residence times but also a contribution from the atomization step. With the coated tubes, the tendency is less for the atomization to extend much beyond the absorbance maximum. The increased porpity of the uncoated tubes causes a greater portion of the analyte atomization to continue beyond the peak absorbance. Tessari and Torsi (19) have also provided evidence for this effect. As the temperature increases, some analyte diffuses into the graphite tube, the direction of this diffusion being reversed with the depletion of the analyte surface layer. Sensitivity. The absolute sensitivities, for both the peak and integration modes of measurement, are listed in Table V. The sensitivities obtained by the strip-chart recorder were measured with a system time constant of 240 ms, those by the integration and oscilloscopic peak methods with a time constant of 18 ms ( 5 ) .For reasons explained in previous papers ( 5 , 6 ) ,the peak sensitivities recorded with the strip-chart recorder are generally lower than the oscilloscopic peak sensitivities, the difference between the two values being a function of the volatility of the analyte. The integral sensitivity is directly dependent upon the magnitude of 7 2 ( I ) . Under the conditions attained with the HGA-2100 ( T ] / T ~N 1,Table IV), those elements exhibiting 7 2 values > 1.0 s will, to a first approximation, give greater sensitivity by the integration mode. Conversely, those elements having 7 2 values < 1.0 s will give greater absolute sensitivity with the peak mode of measurement. These conclusions are evident from the data presented in Tables IV and V. The single exception is Mo, for which a T~ value of 0.85 s leads to an integral sensitivity 50??larger than that of the peak. Table V shows that the sensitivities obtained with the coated tubes are comparable to or better than those obtained with the uncoated tubes. This is a consequence of the smaller 71 values obtained with the coated tubes. According to L'vov's model of atomization ( I , 5 ) , the integral absorbance Q, is, to a rough approximation related to the residence time, 72, and the number of atoms in the sample, No, by:
Q = No72 Under the conditions of pulse vaporization 96
(2)
