Flameless Atomic Absorption Spectrometric Determination of Ultratrace Elements in Silicon Tetrachloride T. Y. Kometani Bell Laboratories, 600 Mountain Avenue, Murray Hill, New Jersey 07974
Nanogram per gram levels of Fe, Ni, Cr, Cu, Co, Mn, and Zn in silicon tetrachloride (SiCi4) have been determined quantitatively by flameless atomlc absorption spectrometry with a carbon rod atomizer (CRA). Various procedures for preparing samples prior to analysis and for preventing contamination errors are described in detail. The most effective sampling procedure was evaporation of SICi4 at low temperature and dissolution of the residue in high purity hydrofluoric acid. Losses of trace elements and interferences from inter-elementalor chemical effects in the CRA were negligible. The precisions and accuracies of the method for the seven metals based on the analysis of a silica standard, ranged from f1.4 to fl3% and f4.5 to f18%, respectively.
Ultratrace concentrations of certain transition elements cause unacceptable attenuation of light transmitted through optical glass fibers. Nevertheless, fibers virtually free of absorption losses have been fabricated by introducing modified chemical vapor deposition (MCVD) techniques ( I ) . With this method low-loss fibers are now produced routinely using commercially available reagents, particularly silicon tetrachloride (SiC14). Although monitoring the initial purity of MCVD reagents is not critical for production of acceptable fibers, suitable analytical techniques are required to determine whether impurities are causitive factors for the periodic production of high-loss fibers and to determine the maximum impurity levels that can be tolerated without noticeable optical effects. A procedure has been described previously for the semiquantitative emission spectrometric analysis of SiC14 (2). However, quantitative methods are needed t o evaluate the effects mentioned above. T h e high sensitivity of flameless atomic absorption methods (10 to 100 times more sensitive than conventional flame atomic absorption methods) permits the quantitative determination of ng/g concentrations of metals in solution. This paper describes methods in which a carbon rod atomizer (CRA) is used for the atomic absorption spectrometric determination of Fe, Cu, Ni, Co, Cr, Mn, and Zn. Reliable quantitative data have been obtained by carefully controlling conditions to minimize contamination and interference effects.
EXPERIMENTAL Apparatus. The atomic absorption spectrometer consisted of a Jarrell-Ash 0.5-m monochromator with Varian-Techtron IM-6 amplifier-readout and power supply. A Varian-Techtron Model 63 carbon rod atomizer with pyrolytic graphite coated tubes was used. The life of the tubes was about 100 firings for HF solutions. A Varian-Techtron BC-6 background corrector accessory was used to check for nonspecific absorption. Absorbance peaks were recorded in the peak-read mode in order to avoid errors due to pen lag. An exhaust fan was attached to the rear of the CRA power supply chassis to avoid overheating. Five microliters (pL) of sample solution were injected into the graphite tube with an Excaliber pipet equipped with plastic tips supplied by VarianTechtron. Instrumental conditions for each element are given in Table I.
The choice of voltage/time settings for the dry, ash and atomize stages was dictated by several considerations. 1. During the dry stage the furnace was maintained at -90 "C for a sufficient time interval to completely evaporate the solvent. 2. Ashing was unnecessary in the absence of an organic matrix. However, the ash stage was used to heat the furnace (600 "C) prior to the atomize stage. 3. Conditions for the atomize stage were chosen so that maximum sensitivity was obtained while completely volatilizing the analyte with a single power sequence. Contamination Control. Because (of the low concentration of element being determined (1-500 ng/g), it was necessary to take precautions to minimize contamination errors during all phases of the analysis. Procedures for preventing contamination from labware, reagents, and atmospheric fallout are described below. Disposable polystyrene serological pipets from Falcon were used for volumetric measurements. Sample solutions were prepared in Teflon beakers. Since polystyrene is attacked by Sic&, glass pipets were used for measuring and transferring the sample. All labware was cleaned thoroughly by alternate leachings in hot HC1 (1:l) and hot HN03 (1:l) followed by a thorough distilled water rinse. The plastic injection pipettor tip was a serious source of Fe contamination. HF was found to be more effective then HCl and HN03 in removing Fe from the plastic tips. Tips were cleaned prior to each use by repeatedly drawing in and expelling concentrated HF followed by rinsing with distilled water. Precleaning in this manner allowed each tip to be reused many times without introducing contamination. Aqueous sample solutions were prepared in Teflon beakers and stored in a covered container. Standard solutions were prepared and stored in 15-mL capped polystyrene vials. Acids were purified by double sub-boiling distillation. All rinses and dilutions were made with water purified by demineralization followed by double distillation in a quartz still. The concentration of impurities in acids and water determined by direct injection into the CRA was