Comparison of secondary ion yield data from ... - ACS Publications

sensitivities of the matrix elements Fe and B. Comparison with sensitivity data of the same elements In other matrices. (stainless steel, silicon) sho...
0 downloads 0 Views 701KB Size
290

Anal. Chem. 1982, 54, 290-294

Comparison of Secondary Ion Yield Data from Amorphous and Polycrystalline I ron-Based Alloys M. Riedei,’ H. Gnaser,* and F. 0. Rudenauer” Austrian Research Center, Seibersdorf, Lenaugasse 10, A- 1082 Vienna, Austria

Absolute and relative practical sensitivities for 02+bombardment of 14 elements, present as 3 atom % admixtures in a Fe,,,B,,X, metallic glass matrix, were determlned by SIMS. The variatlon of sensitivity data between elements is similar to that found for pure element samples. The 3 % admixture causes a small but statistlcaiiy significant matrlx effect on the sensitivities of the matrix elements Fe and B. Comparison with sensitivity data of the same elements in other matrices (stainless steel, slilcon) shows that sensitlvities in different matrices are within 30 rei % for most elements indlcatlng the possibility of transferrlng relative sensitivity factor data determined on metallic glasses to other Fe-based alloys and thereby obtalnlng a semiquantltative analysis.

It is well-known that quantitation of the SIMS method by the use of theoretically founded models can only be considered semiquantitative (1). Better analytical results are obtained with empirical algorithms (e.g., the relative sensitivity factor (RSF) method (2-7)). A number of effects have however to be taken into account when applying constant RSF quantitation. First, it was found that alloying an element to a given matrix or changing the matrix composition can change matrix and trace element yields and sensitivities (“matrix effect”) (8-10). An even more complicated situation may arise when a greater number of components is present in the sample or when concentrations are varying over a wide range; in this case relative sensitivity factors of elements generally are mutually interdependent and empirical quantitation algorithms only yield accurate results when standard sample and unknown sample are similar in constitution and concentration (4). Second, it was shown that instrumental effects can drastically influence the results of a quantitative SIMS analysis. Even identical samples, analyzed on instruments of the same type and manufacturer in different laboratories, can produce raw yield data differing by a factor of up to 50 (II), a factor of about 10 remaining even after presumably quantitative correction (12). Two requirements therefore seem to be essential for the relative sensitivity factor method to yield accurate results: (a) a method should be available for tuning a particular SIMS instrument in such a way as to yield reproducible raw data when the sample is analyzed at different times and to correct for small deviations from reproducibility (even more interesting would be a method for tuning different instruments to yield reproducibly similar relative intensity data; fiist results have been obtained (13) using a procedure similar to that described below); (b) standard samples should be available for a wide range of elements and concentrations. Difficulties have however been reported concerning the commonly used standard samples (6,8,9,14). We consider Department of Physical Chemistry and Radiology, Eotvos University, Puskin ut. 11-13, H-1088 Budapest, Hungary. Institute of General Physics, Technical University of Vienna, Karlsplatz 13, A-1040 Vienna, Austria. 0003-2700/82/0354-0290$01.25/0

the metallic glasses to be suitable standards for quantitative SIMS for several reasons: (a) The metallic glasses do not have concentration gradients on the micrometer scale and are single-phase systems. (b) They can be prepared in a broad range of compositions, including normally insoluble systems. (c) Crystallographic orientation effects (15) cannot occur in the isotropic amorphous glasses, an advantage noted by Newbury for the silicon “conventional” glass standards (6). Only a few SIMS investigations of metallic glasses have been reported so far. Blum and co-workers (16) checked the purity of Fe,B1_, type amorphous thin films. Cahn et al. (17) made use of SIMS to determine the diffusion coefficient of B in a Fe-Ni-B glass. Vasilyev et al. (18) have published measurements concerning the difference in ion yield of atomic and cluster ions from the “equilibrium” and the “nonequilibrium” phase of Fe-B alloys. Buhl and Preisinger (19)reported that the SIMS spectrum of amorphous alloys changed during crystallization. All the investigations mentioned were focused on one or a t most two particular alloys. A comparative SIMS study encompassing a larger variety of metallic glasses has not yet been published. It was the aim of this work (a) to provide practical sensitivity and relative sensitivity factor data to be used for quantitative analysis of the Fe-B metallic glass, using well-characterized ternary metallic glasses as calibration standards, (b) to compare these with corresponding data from polycrystalline alloys, and (c) to investigate the matrix effect on the secondary ion emission from binary and ternary metallic glass systems. Part of these results (RSFs of metallic glasses) have already been previously published (20). Chemically, the amorphous metallic systems (metallic glasses or amorphous alloys) contain at least two elements: a metal and a metalloid with a composition close to that of the eutectic. Their most important properties are (a) macroscopic isotropy, (b) spatial compositional homogeneity (Le., random distribution of constituent atoms), (c) metallic-type bonding (in contrast to the glasses in the conventional sense), (d) the experimentally determined coordination number of the atoms generally lies between 11and 13, which represents a packing density similar to that found in the densest crystalline metallic lattices, (e) on a scale exceeding 1-1.5 nm random structure and elemental composition is found. At present, more than 30 alloy systems are known which can form a metallic glass phase under proper conditions. There are several methods for producing the amorphous state from among which the most universally applicable is extremely rapid cooling (> lo6 K/s) of the molten alloy (“superquench”). Further information on structure and applications of metallic glasses can be found in previously published papers (21-23).

EXPERIMENTAL SECTION The experimentswere performed on the Seibersdorf quadrupole ion microprobe, described elsewhere (24,25). The pressure in the target chamber was in the IO-* torr range during the meabeam of 10 keV energy surements. A mass filtered primary 02+ was produced in a differentially pumped duoplasmatron ion source, focused to a diameter of about 20 fim at the target and rastered with normal incidence across an area of 200 X 200 pm for sputter cleaning of the sample; after reaching steady-state 0 1982 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 54, NO. 2, FEBRUARY 1982

conditions (characterized by a constant matrix ion signal) the rastered area was reduced to a size of 100 X 100 pm for the actual measurement. Usually, a surface layer of the order of 100 nm thickness was sputtered off during this precleaning procedure. The primary ion current was measured in a Faraday cup arrangement and was in the 20 nA range. Secondary ions were analyzed in a commercially avaiable energy analyzer/mass filter arrangement (Riber SQ 156) and detected by an open multiplier operated in the counting mode. The instrument was tuned to accept ions with a central energy around 50 eV and a band-pass of 15 eV width. Secondary ion yield measurements were performed on two analytical areas (100 X 100 Mm) of each sample. Three limited-rangemass spectra of singly charged positive atomic ions were recorded in each analytical area, covering the isotopic range of boron, iron, and the X-element respectively; the isotope peak heights were automatically extracted from the mass scans and appropriately averaged for each set of spectra. Instrument control and data acquisition were fully computerized using a stand-alonePDP 11/34 minicomputer and a CAMAC 1/0system (25). The elemental compositions of the two series of metallic glass alloys studied were: (a) Fe-B with 11.7, 16.6, and 21.6 atom % B content; (b) FewB1,X3,where X stands for a transition metal of the 3d, 4d, and 5d periods (V, Cr, Mn, Co, Ni; Nb, Mo, Ru, Rh, Pd; Ta, W, Os, Ir). The concentration of the X element was generally 3 atom % with the exception of V (3.8 atom %) and Mn (1.5 atom %). The samples have been prepared by the Central Research Institute of Physics of the Hungarian Academy of Sciences, using the method of rapid quenching from molten alloys of high purity, the cooling rate was about lo6 K/s. The composition of the samples was checked by atomic absorption spectrometry (26) and the homogeneity by electron microprobe measurements. Our SIMS measurements proved that the concentration of impurities is generally below the detection limit (10-50 ppm). The amorphous state of the samples has been verified by X-ray diffraction methods. The samples were in the shape of long strips of approximately 1 mm width and 25-30 pm thickness.

DATA EVALUATION For a given element A in a given matrix the practical sensitivity &,(A) (=detected secondary current/(primary current x concentration)) generally depends strongly on the instrument type (the total instrumental transmission factor entering into the detected secondary ion current). The relative sensitivity factors RSF(A) principally may become independent of the instrument type, since, by proper choice of instrument tuning (including pre- and postanalyzer focusing, energy band-pass, extraction angle, etc.) the transmission of one instrument can be made approximately proportional to that of another instrument (at least for a limited mass range) when the above mentioned instrument parameters are identical. In the context of this work, absolute and relative practical sensitivities of a great number of elements, differing in mass number by a factor of almost 20 have to be compared. The data were accumulated during a measuring period extending over more than 2 weeks so that special precautions had to be taken to avoid the problems connected with instrument drift and reproducibility of measurements. It was observed during the measurements that sensitivity values critically depend on the adjustment of instrument parameters. A method therefore had to be developed which allows one to verify that the instrument is adjusted the same way as during the original measurements of a “primary calibration sample” and which allows for correction for small deviations in adjustment. In this work we have used a phenomenologic standardization procedure which we have called the “PCS (primary calibration standard) method”: (1) A ”primary calibration standard (PCS)” is chosen; this should be a well-characterized homogeneous sample containing elements with mass numbers spread out over a large mass range. The metallic glms sample Fe&,,Ws from the Fe-B-X system described above fulfills these requirements. The wide

291

Table I. Standard Operating Conditions (SOC) for Fe-B-X Metallic Glass System (Primary Beam and Sample Related Parameters) primary ions primary ion current beam diameter angle of incidence residual gas pressure in sample chamber oxygen bleeding presputter raster analytical raster secondary ion takeoff

10 keV, W2+,mass filtered ca. 20 nA ca. 50 pm 0” with respect to target normal