Introduction

The central idea in the atomic absorption spectrometric determination of selenium in a graphite furnace (GF) after a hydride generation (HG) is the coating of the graphite tubes with iridium which is able to enrich selenium quantitatively. This opens lower determination limits than the classical hydride-AAS.
The today's usual technology of a hydride generation unites the sample and reagents via a flow-injektion (FI). Besides the older batch-system lasts. Both techniques have strenghts, the first offers automatizableness and the use of very small sample volumes and the second a better adapted process regarding the chemical kinetics of the hydride generation and the use of relatively large sample volumes. At present the market offers FI- and Batch-HG-AA-spectrometers with a classical atomization via H₂Se and FI-HG-GF-AA-spectrometers with an atomization via Ir_xSe_y.
Up to now also the LUFA Kassel used the classical FI-HG-AAS for  the analysis of selenium in feed and plants. Particularly for feed the analysis often is not simple and requires evaluations via standard additions. As the determination limits often don't permit to minimize matrix problems via dilutions or to determinate natural concentrations in plants, the more detection sensitive HG-GF coupling was adapted using already available commercial measuring instruments and a self-designed component and tested in the comparison with the FI-HG-AAS.

HG-GF coupling

A PE FIAS 400 functions as hydride generator, a PE 4100 ZL as AA-spectrometer. The coupling is directly possible after adjustment of the temporal operational sequences.
Normally the interface is the gas-liquid-separator of the FIAS. Here however the coupling was modified substantielly.
Because of volume problems with the commercially available separator already in the classical FI-mode of operation, leading already in former times to the construction of a separator with larger volume, and the positive experiences with this self-designed separator as a batch-reactor in the mercury-AAS with amalgame technique, a new type of batch-reactor (see pict.1) was created and placed on the interface to the spectrometer. The prototype with a screw cap was manufactured from acrylglass (Plexiglas^®) and has a content of 400 ml.

In this configuration the FI-system is only responsible for the automatical transport of the sample, reagents and  carrier gas, the hydogenation and the gas-liquid-separation take place in the discontinuously working batch-reactor. With this FI-HG-GF-AAS, which exactly is an automated Batch-HG-GF-AAS, it succeeded in uniting the more detection sensitive HG-GF-AAS compared with the HG-AAS and the above-mentioned strengths of the FI- and batch-techniques.

Pict.1 Batch-reactor
 

Analytical chemistry and statistical parameters

Sample preparation: microwave- (accelerated-solvent-) extraction (0.5 g sample, 3 ml HNO₃ (w = 65 %), 2 ml H₂O₂ (w = 30 %); 4 min: 250 W / 2 min: 0 W / 1.5 min: 250 W / 2 min: 400 W / 5 min: 850 W / 1 min: 250 W), final volume of the extract (after fill up with H₂O): 25 ml

Prereduction: 2.5 ml extract, 2 ml HCl (w = see below), water bath (temperature/time: see below), final volume of the measurement solution (after fill up with H₂O): 10 ml
HCl-concentration variations: w(HCl) = a) 25 %, b) 32 %, c) 37 %
variations of heating: d) 70 °C / 60 min, e) 80 °C / 25 min
suitable: bd, cd, be, ce; favored because of safety and need of time: be
less suitable, because the suitability is dependent on the sample matrix: ad, ae

Measurement FI-HG-GF-AAS: equipment: PE FIAS 400 with batch-reactor (90 sec gas outlet/GF-enrichment stage), Perkin Elmer AAS 4100 ZL (platform coated with Ir) (atomization temperature: 2000°C))
calibration: (0.5-10 µg/l, 200 µl sampling loop, reagents like FI-HG-AAS): linear (y = 1.064x - 0.020), r² = 0.9990, s(V_r) = 2.2 %
DL (calculated from the calibration): 0.5 µg/l; DL (calculated via the blank value method): 0.2 µg/l
calibration: (0.05-1 µg/ , 2 ml-sampling loop, reagents like FI-HG-AAS): linear (y = 1.002x - 0.013), r² = 0.9991 , s(V_r) = 1.9 %
DL (calculated from the calibration): 0.05 µg/l; DL (calculated via the blank value method): 0.02 µg/l
type of measurement: for feed and plants: compared with aqueous, acid matrix adapted recalibration standards
 

Pict.2  FI-HG-GF-AAS

Results

A validating of the suitability of the FI-HG-GF-AAS took place via 2 NIST-standards (SRM 1515 apple leaves, SRM 1568A rice flour) in 4 parallels each.
 

Tab.1 / Results of standard reference materials

The criterion for a sufficiently exact determination (-u<m-µ<+u) with the secondary condition (s<u) was fulfilled. With SRM 1515 also a successful validating of the constance of the method was carried out during two years (n = 14, m = 0.050 mg/kg, s = 0.003 mg/kg).

For congruent ranges of measurement a testing of the equivalence to the FI-HG-AAS took place by means of a mean value differences test with 20 samples of feed (fourfold determination) and 15 samples of grass out of Se-fertilizing tests (sixfold determination). In part (7 feed) the table 2 shows the results (m = mean content, s = standard deviation, rr = recovery rate of the standard addition).
 

Tab. 2 / Determinations of selenium in feed  a) FI-HG-AAS / b) FI-HG-GF-AAS

Testing of the equivalence of the methods a and b by means of a mean value differences test (n = 20):

1. mean value (x_D) of the differences of the mean contents (column 2 minus 5) = 0.025 mg/kg

2. standard deviation (s_D) of the differences of the mean contents = 0.093 mg/kg

3. testing by means of a t-test, whether x_D is significantly different from zero:

testing value: t_P = Ix_DI / s_D • n^1/2 = 1.20 / comparative value: t_V (P = 95%) = 2.09

t_P < t_V, i.e. x_D is not significantly different from zero,
i.e. the methods a and b are equivalent !
 

Thus above 0.2 mg/kg (1 µg/L) the FI-HG-GF-AAS and the FI-HG-AAS are signifcantly equivalent - also confirmed by the the grass samples. The within-run precisions of the FI-HG-GF-AAS (as a rule <5 %) proved to be easily better than those of the FI-HG-AAS (as a rule <10 %).
Good within-run precisions of as a rule <10 % the FI-HG-GF-AAS also showed below the determination limit of the FI-HG-AAS (0.04 mg/kg (0.2 µg/l)). This was found out with grass samples of normal contents as well as dilutions. For the accuracy <0.2 µg/l it can be stated that comparisons undiluted / diluted corresponded well. A matrix-adapted, certified standard is still looked for.

All the feed and grass analyes first were evaluated via standard additions. For the grasses this was proved as unnecessary in both measurement techniques, the recovery rates were at approximately 100 %. The columns 4 and 7 of table 2 show that for feed only the FI-HG-GF-AAS offers this, a great practical advantage which the chemical kinetics of the batch-mode make possible.

The redox potentials between the selenium oxidation states and the pair 2H^(-)/H₂ on the other hand are not to be influenced. Supplementary experiments with selenite and selenate salts resulted in this. Without a prereduction the both measurement techniques didn't find selenium for both salts after a digestion and for the latter after a pure dissolving.
A complete recovery there was in all cases after a prereduction.

Summary

For the selenium analysis the LUFA Kassel constructed a new variant of a FI-HG-GF-AAS, which also can be described as an automated Batch-HG-GF-AAS, using a flow-injection automat for hydride generation, a self-produced batch-reactor and a graphite furnace atomic absorption spectrometer.

The experimental experiences permit the following central statements: