Analysis of High Temperature Alloys by X-Ray Fluorescence RUBEN M. BRISSEY Thomson Laboratory, General Electric Co., Lynn, Mass. A portion of the secondary or fluorescent radiation falling in a limited sector of the horizontal plane passes through a short collimator auto a mica crystal. This analyzer is of the bent crystal type, so designed that the 337 planes are in the R position. The radiation passes through the crystal and is diffracted to give focusing a t angles satisfying Bragg’s equation:
OMPLEXITY of alloys for high temperature applications has added much to the problems of the analytical chemist. Any method of analysis by which compositions may he determined without separating the alloy into its component elements is attractive to those who must obtain analysis data on either production stock or experimental alloy heats. One method fulfilling this requirement n-hich has created considerable interest is aualysis by x-ray fluorescence. The alloy sample is subjected to x-radiation of sufficient energy to excite the characteristic radiation of the elements in the sample. The intensity of these characteristics lines should he a function of concentration. The work in x-ray analysis up to 1932 has been well summarized by von Hevesy ( I ) . The early experiments, where secondary x-radiation was measured, used film techniques for detection, with ratios between line intensities used to obtain compositions in much the same manner used in spectrogr%phicanalysis. Indeed, spectrographic methods were and are being applied t o radiation in the x-ray region. Within the past few years, instruments capable of producing, resolving, and measuring the intensity of characteristic radiation have been marketed commercially. These instruments use Geiger counters to detect the measure intensity.
nh = 2d sin R The radiation passes through a 0.3’slit and enters the Geiger counter tube through a beryllium window. Suitable electronic circuits provide far chart recording of location and relative intensity of characteristic lines. The recorder response to intensity is logmathmic. A chart recarding is shown in Figure 2. The numbers appearing above selected peaks in the figure indicate an approximate line intensity expressed a8 counts per second. This automatic scanning is used if only qualitative answers are desired. For quantitative work, this recorder is not used. Instead, the goniometer is set at the angle of maximum intensity for the line in question. Time required for a preset number of counts is recorded by a timer. The total intensity may then be expressed &s counts per second. This intensity results from chitrac-
OPERATION O F ANALYZER
The instrument used for the work described in this paper was a General Electric X R D 3 S unit, pictured in Figure 1.
Samples should have a flat surface that covers the sample holder window. A belt sander is frequently used to smooth the sample if, as cut from stock, it has an irregular surface. A flat area of the metallic sample is exposed to x-rays from a tube containing a tungsten target. The tube is operated a t a peak voltage of approximately 50 kv. This voltage is adequate to praduee a continuous spectrum capable of exciting the X radiation of all elements considered in this particular alloy type. The tube amperage is maintained near the maximum rating for the tube, which in this case is 50 ma. A maximum intensity Figure 1. X-Ray Fluorescent Analyzer of the fluorescent radiation is obtained a t the highest tube amperage. However, very high oounting rates introduce errors due to coincident counts. Since no mare than 16,384 counts can be timed ou this instrument, significant errors in timing can also result if the interval is short. T o avoid these errors, the tube amperage may be reduced when an element’s f l u o r e s c e n t i n t e n s i t y app r o a c h e s 2000 c o u n t s p e r second a t 50 ma. Tungsten is used as the target material, since tubes containing it are available that will withstand the operating conditions. Scattered characteristic radiation from this tube that reaches the counter doesnot interfere in the analysis, as would result if a molybdenum target were used for molybdenum Fignure 2. C h a r t Recording of Characteristic X-Ray Lines from a Typical High analyses. Temperature Alloy 190
191
V O L U M E 25, NO. 1, J A N U A R Y 1 9 5 3 Table I. Summary of Data from Homogeneity Samples Intensity, Counts/Second from K, Line Sample S o .
h-i
.\IO
84.5
1 2 3 4 5 6
1678 1670 1665 1710 1719 1714 1728 1697 1728 1696 1728 1665 1700.5 3.79
85.8 85.8 84.5 83.6 85.1 85.8 85.1 86.1 86.1 86 1 83.6 85.24 2,99 0.835
7
8 9 10
Highest ralue Loaest value Average Max. deviation, S Standarddeviation
23.2
co
Fe
277.6 274.0 277.3 276.3 275.7 274.0 277.0 275.5 277.6 274.3 277.6 274.0 275.9 1.31
56.75 56.3
1.46
Cr 101.2 100.3 99.1 101.1 99.1 100.1 99.2 101,s 100.7 100.7 101.5 99.1 100.3 2.42
5 5 85
56.0 56.6 57.1 57.2 57.4 57.8 57.0 57.8 55.85 56.8 3.49 0.537
An operator has determined the 5 elements in 20 samples in 1 day. The intensity arising from a given atomic per cent of an element will be dependent upon the other elements and their relative concentrations in the sample. This is essentially due to absorption effects. Changing the concentration of an element with a high absorption for a line used in intensity measurements will alter the intensity of the line, although the concentration of the emitting element is unchanged. This absorption results in excitation of the absorbing element. If the lines excited by the absorption include those subjected to intensity measurements, the values obtained \\-ill be influenced by the concentration of the element emitting the highly absorbed line.
0.905
REFERENCE SAMPLES
Composition, Wt. % Chemical analysis
9 60
-
53.64
10.32
2.58
18.17
Table 11. Comparison of Intensities from Samples at Different Stages of RIetallurgical Processing Element Molybdenum h'ickel Cobalt Iron
Saniple 1 2 1 2
128 6 126.8 1427 1443
1 2 2
636 642 174.7 171.2
1 2
189.4 190 5
1
Chromium 0
Intensity, Counts/Seconda Cast Forged
Sample area exposed, 31,
x
3/r
49.7 ma.. except h-1run at 24.1 ma.
~liernical~ n a l y s i s , Kt. %
126 1 126.3 1457 1444 639 659 172.8 172.1
7.47 7.83
... ...
13.64 13.55 4.68 4.48
1894 18 98 inch; peak voltage, 49.7 h v , ; current, 198.8 198.9
teiistic radiation from the element in the alloy plus a background intensitv. The latter consists of some scattered radiation from the tube and external radiation that activates the counter. Subtracting background gives an intensity resulting only from the element in the sample. When a series of analyses is made, all data for one eIement in all unknowns and reference samples are taken without shifting the goniometer. Including reference samples a t the time analysis data are taken automatically compensates for intensity changes caused by changing parts of the instrument of geometric differences due to the method of assembly. -4s a rule, four time readings for a given number of counts are recorded and averaged for each line. Background intensities are obtained a t angles free from characteristic radiation and near each line measured. Once determined for an alloy type, this background is used without rechecking until some unit of the instrument is replaced. Unless the counting rate is very low, a total of at least 16,384 counts is recorded. This gives a probable counting error of =!=0.5%.
Since the composition of alloys run on a routine basis is known within reasonable limits! the absorption problem has been avoided by using chemically analyzed reference samples with x-ray absorption properties mat rhing the unknowns. Similar abeorp tion properties are best assured hy using a similar composition. Surh references have been used in the analysis of high temperature alloys for chromium, cobalt, iron, molybdenum, and nickel. An early check was made on this alloy type to determine whether excessive sample variation would be encountered and to establish t,he influence of metallurgical processing. As a check on sample variation, int,ensity measurements were made for the 5 elements in 10 samples from a single production lot. These data are Q h o min Table I. The area of the sample Peak voltage was 49.7 kv. and cur' 1 2 x ' 1 2 inch. e-vsed rent 49.7 ma. These values are corrected for background and are for the I