1706
Anal. Chem. 1980, 52, 1706-1708
Sub-Part-per-Billion Iron Determination by Laser Intracavity Absorption Spectrometry T. D. Harris and J. W. Mitchell" Bell Laboratories, Murray Hill, New Jersey 07974
(15) more recently obtained the most sensitive absorption detection of sodium yet achieved, 5 x lo5 atoms/cm3, using a 15-cm intracavity cell and photoelectric detection of dye laser quenching of the D2 line. Recently, G. H. Atkinson and coworkers (16) performed a thorough study of the spectroscopic fidelity and quantitative aspects of the intracavity laser detection of NO2. Reproducible spectra were obtained at a given pressure and linearity was seen over the pressure range 10 to 300 mTorr. Although these reports have clearly demonstrated the potential of the intracavity method for analytical use, most of them have simply illustrated the sensitivity of the technique by selecting artificial samples amenable to study with an intracavity spectrometer. T o our knowledge, no previous publication has yet described the application of the intracavity technique for quantitative determinations of trace species present in real samples. It is our purpose in this paper to demonstrate that ordinary spectrophotometric procedures now generally applied in analyses can be exploited for high sensitivity quantitative determinations via laser intracavity spectrophotometry. Conventional spectrophotometric reactions may be combined with laser intracavity absorption measurements and careful sample processing to provide a n extremely sensitive and reasonably selective way of quantitively determining a single element with high reliability. For several reasons, the determination of iron was undertaken. The measurement of this element at ultratrace levels continues to be important (17). It is the single most prevalent and detrimental transition element impurity in optical waveguide grade SiC1,. Reliable quantitative determination below the nanogram level continues to be particularly difficult because of blank problems and restricted measurement limits of existing methods. For example neutron (30-day irradiation a t 1013 n/cmz s) and proton activations (13 MeV, 5 PA, 5 h) detect only 50 and 15 ng, respectively, in the irradiated sample. Atomic absorption spectrometry and stable and radioisotope dilution methods have respective measurement limits in special samples of 1, 3, and 12 ng/mL. The laser intracavity technique in principle could be used to obtain several orders of magnitude improvement in measurement limits. In the case of iron determination, the ferrozine reaction (18, 19) was selected because of its high absorption coefficient (28 600), high formation constant ( - 3 x and its aqueous solubility. T h e chemistry involved in the formation of the iron(I1) chromophore is also sufficiently complex to be quite representative of the general chemical problems that would arise during the ultratrace determination of an element by spectrophotometry.
Laser intracavity absorption has been applied to the quantitative spectrophotometric determination of broad band absorbing solution species, one of the most prevalent types of samples in trace elemental analysis. Chemical reactions and sample processing procedures used for conventional spectrophotometry of iron with ferrozine have been applied to the measurement of the element in the range 0.5 to 3.0 ng/mL. Procedures for obtaining low and reproducible blanks, 0.45 f 0.09 ng, by using specially ultrapurified analytical reagents are described. Determinations of iron in silicon tetrachloride are reported.
The increased sensitivity of the laser intracavity absorption technique was demonstrated experimentally in 1971-72 (1-3). Since these initial reports, the enhanced sensitivity of the method compared to that of conventional absorption has been well documented. Enhancements of lo3 have been observed for pure Cs vapors ( 4 ) , lo3 to lo5 for I2 ( 5 , 6 ) , several orders of magnitude for HCO and NH2 radicals, respectively and 10' for NOz a t 1900.04 cm-' (8). The ability to detect weak absorbers in solution has been demonstrated for Eu(NO,), in methanol (9) and the qualitative detection of species in flames after injection of solutions was demonstrated for Ba+ and Sr in a n air-acetylene flame (2). Quantitative investigations of the dependence of the intracavity effect on concentration of the sample have included primarily examinations of rare earths and sodium in aqueous solutions or in flames. Spiker and Shirk (10) showed empirical relationships between laser intensity and Ho3+ or Pr3+ conto M in aqueous solucentrations in the range 5 x tions. Other investigations of Eu3+and Pr3+ in methanol and water, respectively, showed linear relationships of apparent intracavity absorbance vs. rare earth concentration over the range 0.048 t o 0.24 and 0.0012 to 0.006 M, respectively (11). A small enhancement factor of 35 was observed. Intracavity detection of sodium has been investigated repeatedly. Sodium initially present a t the 2 x lo-'' ng/mL level in an aqueous solution was detected in an intracavity air-natural gas flame (12). Green and Latz (13) reported that sodium a t the 1 ng/mL level could be easily detected qualitatively in 10% HzO-90% ethanol standard solutions. However, limitations of densitometric detection restricted construction of calibration curves to the 1-to 100-ppm range. Approximately linear laser output signals with Eu3+ concentrations (2000 to 3000 ppm) in methanol were also reported by these authors (13). Quantitative aspects of the intracavity detection of atomic absorption of trace elements in a flame were investigated more thoroughly by Maeda, et al. (14). Na, Li, Sr, Ba, and Cs atoms in the wavelength range from 450 to 680 nm were detected by a dye laser with an air-acetylene flame inside the cavity. For sodium and lithium, analytical curves were measured, and sensitivity and detection limits were reported. Respective detection limits of 0.2 and 5 ng/mL were found. Sensitivity improvements of 2500 were obtained for Na over conventional atomic absorption. However. the detection limit is only slightly improved due to fluctuations in the detected signals resulting from instabilities of the laser system. Maeda et al.
(a,
0003-2700/80/0352-1706$01 O O i O
EXPERIMENTAL Reagents and Apparatus. Design parameters and operation of the laser spectrometer are described in detail elsewhere (20). ,411 stock solutions were prepared in 100-mL polypropylene volumetric flasks (SGA Scientific) and samples were prepared in 25-mL polypropylene volumetric flasks (Bel Art Products, Pequannock, N.J.). Reagent transfers were accomplished using 1-mL plastic serological pipets (Falcon Plastics). Silicon tetrachloride samples were prepared in 50-mL Teflon beakers (SGA Scientific). Ferrozine was purchased from Hach Chemical, Ames, Iowa, and used without further purification. Hydrazine (Eastman C
1980
American Chemical Society
A N A L Y T I C A L CHEMISTRY, VOL. 52,
Chemicals, 95%) and acetic acid (Apache Chemical, high purity) were purified by low temperature sublimation, frozen, and stored in Teflon bottles before use (21). Ultrex ammonium hydroxide was purchased from J. T. Baker Chemical Co. and used as received. Standard iron samples were prepared from Reagent ACS Fe(NHJ2SO46H20and standardized both spectrophotometrically and by EDTA titration. All acids used in sample or reagent preparation except phosphoric were high purity doubly sub-boiling distilled, purchased from the National Bureau of Standards, Washington, D.C. Phosphoric acid was prepared by hydrolysis of Ultrex P205purchased from J. T. Baker Chemical Co. All acids used to leach or rinse labware were low sodium M.O.S. Electronic Grade from J. T. Baker Chemical Co. Water used throughout was demineralized and distilled twice in a quartz apparatus. Procedure. All plastic labware except pipets were prepared by soaking for 24 h in 8 M HNOBat 80 "C, and then in 6 M HC1 for 24 h at 80 "C. Prior to each use, each item was rinsed twice with concentrated H F and then 5 times with distilled water. Pipets were cleaned by filling and draining with concentrated HF four times and then repeating the procedure with distilled water. It is essential that the graduation markings of the pipets not be allowed to come in contact with acids of the sample solution since the ink pigments contain metallic impurities. All reagent and sample preparation and manipulation except introduction of the sample into the cell were carried out under an all-plastic laminar flow hood located in a Class 1000 clean room. Ferrozine solution was prepared by dissolving 0.5 g of the solid in 50 mL of distilled water, adding HC1, and diluting with water such that the final volume was 100 mL and the acid concentration approximately 0.5 M . Ammonium acetate buffer was prepared by partial neutralization of glacial acetic acid with ammonium hydroxide solution t o give a final pH of 4.0. Preparation and handling of hydrazine were as previously reported (21). It was used as either the concentrate diluted with water (1:2) or neutralized to pH 2 with HC1 or HKO+ This neutralization required about three parts acid to one part hydrazine yielding a final hydrazine concentration of approximately 8 M. Solutions of anions for interference tests were prepared by partial neutralization of the appropriate acid (hydrofluoric, phosphoric, etc.) with ammonia. Cations were prepared from the native pure metal, dissolved in nitric acid and diluted. A sample, blank, or standard was prepared according to the following procedure. To a 25-mL volumetric flask were added 0.20 mL ferrozine followed by 0 20 to 2.00 mL hydrazine. The hydrazine concentration was dictated by the level of interfering anion in the sample. A t this point the solution must be highly acidic, preferably with a pH of 1.5 or less. After these two additions, the sample or the appropriate volume of iron solution with or without interfering ions mas added. The volume of this addition may vary from 0.10 to 10.0 mL. The final addition is 0.50 to 1.50 mL of ammonium acetate. The samples are then diluted to volume and mixed. Absorbance measurements may be taken from 2 min to 24 h later.
RESULTS A N D DISCUSSION Reagent Selection and Purification. For the purposes of chemical analyses a t or below one per billion, concentrated reagents with impurity levels a few parts per billion or less are necessary. Since the majority of analytical samples are at one point or another in aqueous solution, water is the solvent of choice, and in ultrapurified water the iron blank is exceedingly low. Ferrozine reagent possesses the most favorable combination of aqueous solubility, absorption coefficient, and stability constant. Hydrazine was chosen as the reducing agent for several reasons. First, it can be purified by sublimation as noted earlier. Second, hydrazine has a large reduction potential, which minimizes the concentration needed t o accomplish reduction of iron. Third, its oxidation and decomposition products, nitrogen, ammonia, and water, are highly unlikely to cause chemical interferences. Fourth, its purification leads to a highly concentrated product, again minimizing the contribution to blank iron levels. Acetic acid and ammonia were selected as buffer agents because of the efficiency of their purification by cryogenic sublimation (22)
NO. 11, SEPTEMBER 1980
1707
Table I. Repeatability of Laser Intracavity Absorption Measurementsa Fe added, ngimL blank 0.5 1.0 1.5 2.0
absorbance x
lo3
I
I.[
I11
1.8 3.7 6.2 7.1 9.1
2.8
3.1
6.2 7.8 9.1
6.4 7.3 9.4
1-cm cells, absorbance measured a t 584 nm, each series of standard solutions was prepared separately a t three-week intervals and then measured. a
and isopiestic distillation, respectively (23). The success of an ultratrace analytical scheme is determined in large part by its simplicity. A simple procedure if properly executed will invariably yield lower blanks, and hence lower detection limits, than a complex procedure. T h e strategy of the present work was to prepare samples and standards in the fewest steps from the purest possible materials. Blank Determinations and Calibration Plots. T o test the variability of the absorbance blank, fourteen measurements were made on seven solutions. The experimental result, (3.8 h 0.2) x lo4, shows that low and reproducible absorbance blanks were measured quantitively via laser intracavity absorption spectrophotometry. This absorbance blank value corresponds to 0.45 f 0.09 ppb of iron, a particularly low and reproducible blank for a chemical procedure involving this ubiquitous element. T h e 2 0 limit of 4~0.09ppb can be considered the limit below which iron could not be detected in the presence of the blank. The predominant contribution t o the blank resulted from the ferrozine reagent itself, which as the solid contains - 2 ppm of iron. More than 95% of the blank value originates from this reagent. A carefully controlled direct route for synthesis of this reagent might allow further purification and reduction of the blank. Linear calibration curves were constructed reproducibly and corresponding data are reported in Table I. The repeatability of these data obtained independently and on different occasions separated by periods of several weeks clearly demonstrates that the chemical production of the ferrozine complex is sufficiently complete and stable for quantitative applications. These data also further document that our laser spectrometer (20) is suitable for making quantitative absorption measurements reproducibly. Interferences. Spectral interference from cationic species is expected from Co, Ni, and Cu(1) (19). Since iron is usually the most difficult of the transition metals t o eliminate, one would not expect to find significantly more of these species than iron in an ultrapure material. With iron a t the 2 ppb level, a 10-fold excess of Co or Ni did not introduce any detectable error. The measured absorbance of Fe remained stable a t 1.0 x in the presence of these impurities. Cu present a t this level interferes significantly as shown by the increase of the absorbance t o 1.9 x Anionic interferences are potentially more serious. No common inorganic anion forms a stronger ferrous complex in acidic solution than the ferrozine anion. However, fluoride and phosphate form very strong ferric complexes. T h e resultant stabilization of the ferric species by fluoride or M effectively prevented phosphate concentrations of reduction of any iron by normal hydrazine concentrations. Reduction was accomplished in the presence of either anion a t lo-*M by raising the hydrazine concentration t o 7.5 M during the reduction step. Calibration curves virtually identical t o those resulting from date in Table I could be M phosphate or constructed in the presence of 1 x fluoride. This is particularly noteworthy considering that the
Anal. Chem. 1980, 52, 1708-1710
1708
total salt concentration of the measured sample exceeded 1.5 M. Applications. The purity of silicon tetrachloride, a starting reagent for the production of epitaxial silicon films and optical waveguide materials, must be assured with respect to transition metal impurities. Iron has been previously identified as usually the only detectable, soluble transition metal impurity present in the prefiltered material (24). Since high purity silicon tetrachloride may contain sub-nanogram amounts of this impurity, analysis by laser intracavity spectrophotometry was performed as a real world application. Four aliquots of a purified sample were added to precleaned 100-mL Teflon beakers and evaporated to dryness on a ceramic top hot plate located inside a laminar flow hood located in a clean room. Samples were vaporized within 5 min. Four empty beakers were placed along with beakers containing samples and processed identically as blanks. After evaporations of SiCl4, during which iron traces are retained without loss (24),residues were dissolved in 0.5 mL of concentrated H F and the resulting solution was evaporated to dryness. After the addition of 5.0 mL of 1.1M HC1 purified by subboiling distillation, samples were heated briefly and then processed as described in the Experimental section. Quadruplicate determinations yielded blank values of 0.5 & 0.1 ng and iron was found at the 2.3 i 0.2 ppb level in silicon tetrachloride. LITERATURE CITED (1) Peterson, N. C.; Kurylo, M. J.; Braun, W.; Bass, A M.; Keller, R. A. J . Opt. SOC A m . 1971, 6 1 , 746-750.
Thrash, R. J.; Weyssenhoff. H.: Shirk, J. S. J , Chem. Phvs. 1971, 55, 4659-4660 Klein, M B Opt Commun 1972, 5 , 114-116 Childs, W J Fred, M S I Goodman, L S Appl Opt 1974 73 2297-2299 Hansch, T. W.; Schawiow, A. L.; Toschek, P. E. I€€€ J . Quantum Nectron. 1972, 8 , 802-804. Keller, R . A.; Simmons, J. D.; Jennings. D. A. J . Opt. SOC.Am. 1973, 63, 1552-1555. Atkinson, G. H.; Laufer, A. H.; Kurylo, M. J. J . Chem. Phys. 1973, 5 9 , 350-354. Chackerion, C., Jr.; Weisbach, M. F. J . Opt. SOC. A m . 1973, 63, 342-345. Keller, R . A.; Zalewski, E. F; Peterson, N. C. J , Opt. SOC.Am. 1972, 62, 319-326. Spiker, R. C., Jr.; Shirk, J. S. Anal. Chem. 1974, 4 6 , 572-574. Horiick, G.; Codding, E. G. Anal. Chem. 1974, 4 6 , 133-136. Konjevic, R.; Konjevic, N. Spectrosc. Leff. 1973, 6. 177-181. Green, R. B.; Latz, H. W. Specfrosc. Lett. 1974, 7 , 419-430. Maeda, M.; Ishitsuka, F.; Miyazoe, Y. Opt. Commun. 1975, 73, 314-317. Maeda. M.; Ishitsuka, F.; Matsumoto, M.; Miyazoe, Y . Appl. Opt. 1977, 16, 403-406. Atkinson, G. H.; Heimlich, T. N.; Schuyler, M. W. J . Chem. Phys. 1977, 66, 5005-5012. Fishman, M. J.; Erdmann, D. E. Anal. Chem. 1979, 51, 317R-341R. Kundra, S. K.; Katyal, M.; Singh, R . P. Anal. Chem. 1974. 4 6 , 1605- 1606. Gibbs. C. R . Anal. Chern. 1976, 48, 1197-1201. Shirk, J. S.; Harris, T. D.; Mitchell, J. W. Anal. Chem., preceding paper in this issue Mitchell, J. W.; Harris, T. D.; Blitzer, L. D. Anal. Chem. 1980, 52, 774-776. Mitchell, J. W. Anal. Chem. 1978, 5 0 , 194-196. Zief, M.; Mitchell, J. W. "Contamination Control in Trace Element Analysis"; Wiley-Interscience: New York, 1975; pp 104-106. Kometani, T. Y. Anal. Chem. 1977, 49, 2289-2291.
RECEIVED for review January 25, 1980. Accepted June 9, 1980.
Simultaneous Determination of Zirconium and Hafnium in Solutions by X-ray Fluorescence Spectrometry Enzo Ricci" Analytical Chemistry Division, Oak Ridge National Laboratory, P.O. Box X, Oak Ridge, Tennessee 37830
The importance of Zr and Hf in nuclear reactor research and development is well established. A reliable X-ray fluorescence method (XRF) was developed to determine these two elements simuttaneously in (0.5-1.0 M H2S04)solutions at levels ranging from 0.5 to 200 ppm. Handling of serial samples is enabled by a 90-sample changer and by the simple chemistry involved. Ferric hydroxide is precipitated with ammonia in a test tube. After washing, the Fe(OH), slurry Containing all the Zr( OH), and Hf(OH), is transferred to a 6.4 pm thick Mylar mount where it is quickly dried on a steam bath. These mounts are inexpensive, hold well during drying, and provide a
