Results and Discussion

Experiments with IC-cuvettes

We have found that the mean gas phase temperatures as experienced by lead atoms are similar in a Perkin-Elmer HGA-74 and an ICC when atomization proceeds from the tube wall. However, for platform atomization, the ICC provides higher mean gas phase temperatures (see Table II) because there are virtually no cool regions in the ICC which can contribute to lower values. The results in Table II are consistent with the findings of Falk et al. [1], since with platform atomization the appearance of lead atoms is delayed until a temperature gradient has developed in the HGA. Therefore it is expected that the susceptibility to gas phase interference effects should be similar in both systems for volatile elements atomized from the tube wall and this is illustrated by the results given in Table III.

Table II. Spectroscopic temperatures in side- and end-heated (HGA-74) tubes using lead as the thermometric species.
Set temp./oC Atomization conditionsSpectroscopic temp./ oC
2000Platform1770 1900
2000Platform + modifier17101960

Table III. Recoveries (100 * signal in matrix/signal in H2O) for 0.4 ng lead using wall atomization. No ashing, no modifier used.
MatrixSignal evaluationRecovery/%
20 ug NaClHeight 10097
20 ug CuCl2Height196

For platform atomization, however, theoretical considerations would indicate [4] that the higher gas phase temperatures provided by the ICC should reduce interference effects more efficiently. In order to test this hypothesis a number of elements were determined in the presence of copper nitrate and copper chloride using similar heating rates in both furnaces. Copper nitrate was used as a reference solution to unambiguously study the effect of chlorine. To simplify the comparison no matrix modifiers and no ashing were used. As can be seen in Table IV, the peak area recoveries for the ICC are significantly better than with the HGA, and except for aluminium, they are close to 100 %. Of interest is the fact that comparison with aqueous standards shows small interference effects even though no thermal pretreatment was used. To illustrate the performance of the ICC, Fig. 3 shows signals for manganese in the two matrices. It is noteworthy that, for all elements investigated, double peaks appeared in the presence of chloride. The first peak is probably the result of gas phase dissociation of the metal chloride in agreement with Welz et al. [15] who discussed the reasons for observed double peaks when atomizing antimony in the presence of nickel chloride. From the signal shape in Fig. 3 it is obvious that peak height evaluation is, in this case, useless and therefore only peak area data are given in Table IV.

Fig. 3 Signal traces for 0.2 ng of manganese in the presence of Cu(NO3)2 and CuCl2 matrices using a platform equipped ICC. The background signal is shown by the dotted line.

Table IV Peak area recoveries from a chlorine matrix when using the L’vov platform. No ashing, no modifier used.
HGA-500a,b ICCbICCc
Ag609892, 97d
a Values taken from [14]
b Recovery = (100 * signal in 0.1 % CuCl2/signal in 0.1 % Cu(NO 3)2)
c Recovery = (100 * signal in 0.1 % CuCl2/signal in H2O)
d Tube temperature 2000 oC instead of 1600 oC
e H2 added to purge gas

When using modifiers in combination with the L’vov platform, optimum ashing temperatures and peak area evaluation, matrix interference effects are often small [4], even in end-heated furnaces. However, in determining lead and tellurium in a sodium chloride matrix we found a signal depression when using the HGA-500. This interference could be eliminated by employing the ICC [8].

Further problems with spatial non-isothermality concern the use of wall atomization since there is a risk that a sample which spreads towards the tube ends during pretreatment will be volatilized slower than that sample confined to the tube centre. A comparison of the interference effects resulting from sample spreading in side-heated and GTA tubes is shown in Table V for wall atomization. For all elements investigated the signal was significantly suppressed in 5 % of nitric acid when using normal injection mode with the end-heated GTA-96. This was true even for the volatile elements indicating that the sample was spread to tube regions which did not reach sufficient temperatures for atomization even during ramping. With the GTA-96 autosampler it is possible to overcome spreading by injecting the sample at a controlled rate onto a heated graphite surface; see columns ‘hot injection’ in Table V. The signal suppression for gold must be attributed to effects other than spreading. For the ICC, any effects of spreading are negligible since the peak area recoveries are close to 100 %. The rate of atom formation and hence the peak height might, on the other hand, be influenced by variations in the area of the graphite surface in contact with the analyte, as reflected in greater deviations of the peak height recoveries obtained with the ICC. The determination of vanadium and molybdenum using the end-heated Massmann-type furnace also suffered from tailing and memory effects.

Table V Comparison of interference effects due to sample spreading obtained with side- and end-heated (GTA-96) tubes. Atomization from tube wall, with the GTA-96 using standard pyro-coated tubes.
Element Recoverya,b/%
normal injection
hot injection
Au86 (62)68 (66) 76 (72)
Cd102 (108)49 (76) 99 (84)
Mo97 (103)34 (30) 100 (101)
Pb97 (112)22 (22) 106 (105)
V102 (96)37 (28) 102 (101)
a Recovery = (100 * signal in 5 % HNO3/signal in H2O)
b Peak area and (peak height) results

To give an idea of the sensitivity achievable using the ICC, Table VI lists some peak area characteristic mass data along with those reported by Slavin et al. [2] for the HGA-500. The characteristic mass data for the 17 mm long ICC were recalculated (peak area proportional to l2 where l is the tube length) to correspond to a 28 mm long tube, i.e. the length of an HGA tube. However, atomic vapour might persist outside the open ICC which could result in overestimated sensitivities. Since the cross sectional areas of the ICC (25 mm) and HGA (26.4 mm) are similar it can be seen that the full length of the HGA is not completely utilized indicating losses of atomic vapour at cooler tube ends. It should be mentioned that longer ICCs can be used, but these require larger power supplies than those typically used in current instrumentation to provide comparable heating rates and maximum temperatures.

Table VI Characteristic masses for ICC and HGA-500 [2].
Element Atomization mode Characteristic mass/pg
17 mm
28 mma
28 mm
AgPlatform2.5 0.91.4
AlPlatform21 7.710
MnPlatform3.5 1.32.0
MoWall10.6 3.99.0
PbPlatform19 7.011
SnPlatform47 1723
VWall43 1630
a Recalculated, see text

Previous page
Next page
Return to the Conference Presentations page, or the Atomic Spectrometry Group home page