2. Nickel concentration in cross sections of 6 3. Nickel concentration in longitudinal sections LIST OF FIGURES FIGURE No. PAGE 1. NBS-461 steel polished for inclusion counting 4 2. Typical inclusion in longitudinal section 5 3. Cross section of NBS-461 steel etched for 9 s 7 4. Longitudinal section of NBS steel etched for 7 Metallographic Characterization of an NBS Spectrometric Low-Alloy Steel Standard R. E. Michaelis, H. Yakowitz, and G. A. Moore (January 2, 1964) The spectrometric standard steel designated NBS Low-Alloy Steel 461 was investigated by means of electron probe microanalysis and quantitative metallographic techniques employing a digital computer. Electron probe microanalysis showed the steel to be homogeneous in nickel and iron at two to four microns of spatial resolution. The average of all determinations agreed with the certified values for these elements. Inclusions in the steel were identified, classified as to size and shape, and counted. Mean free path data on the inclusions were calculated. The ASTM ferrite grain size number was deduced as 13.5 for the steel in the unetched condition. From the mean free paths in ferrite and pearlite, it was found that the steel is structurally homogeneous at a five micron level. It is concluded that the homogeneity level corresponds closely to the grain size of the material. It is further concluded that NBS-461 steel is sufficiently homogeneous that any present microanalytical technique can be carried out with little chance of inaccuracy due to inhomogeneity. 1. Introduction Most of the modern rapid instrumental methods of analysis depend on the use of standard samples of composition for calibration. The National Bureau of Standards in cooperation with industrial and government groups, plans, prepares, tests, and certifies selected standards of composition to serve the urgent needs of science and technology in critical areas of calibration. Whereas generally suitable techniques have been developed for the preparation of homogeneous standard samples of composition to be applied in chemical analysis, or in the usual optical emission and x-ray spectrochemical analysis, [1]' some of the newer analytical methods are far more demanding with respect to the requirement of homogeneity. Included in the newer analytical techniques which have stringent requirements for sample homogeneity at the 1 to 50 μ level are those involving the solids mass spectrometer, the laser excitation source in spectrochemical analysis, and the electron-probe microanalyzer. The latter instrument, for example, can perform an analysis with a spot diameter of about 1 μ. An important objective of the spectrochemical standards program is to develop new or improved procedures and methods for preparing, testing, analyzing, and applying the spectrochemical standards. In this, research is directed not only to techniques for providing standards with high measurement excellence, but also to procedures for increasing the applicability of existing and future standards. The purpose of this report is to present the results of a limited but critical evaluation of homogeneity at the 2 to 4 level in one of the available NBS μ Figures in brackets indicate the literature references at the end of this paper. steel spectrochemical standards. The evaluation is intended (1) to provide a more complete characterization of the standard, and (2) to learn of its suitability for application to the newer analytical techniques. The investigation primarily involves a study by the NBS electron probe microanalyzer and by new quantitative metallographic techniques. 2. Selection of Material for Study At present, the NBS does not have available specific standard samples of composition for some of the newer analytical techniques, such as the solids mass spectrometer. However, a number of available NBS standard samples, designed primarily for application in spectrochemical analysis, was prepared and processed to yield material of high homogeneity. Based on the results of extensive testing several years ago by optical emission and x-ray spectrochemical analysis, by chemical analysis, and by metallographic studies, these standard samples were determined to be free from segregation, for the intended applications, for most contained elements. In particular, a set of eight ingot iron and low-alloy steel standards, prepared in the years 1956 and 1957 to contain a graded composition range for about 25 elements, was found to be homogeneous at a level of about 1 mm (1000 μ) for most elements as based on the methods of test then available. It was believed that one of these standards would serve the purpose of this investigation. The study of photomicrographs of these irons and steels indicated structural inhomogeneity for all eight standards at the micron level; however, at least two of the standards appeared to exhibit suitable structural homogeneity at the 10 to 20 μ level. The latter level should be satisfactory for application to either the solids mass spectrometer or to the laser excitation source in optical emission analysis. The material for each standard had been melted | 3. Information Desired and Equipment Used in a one-ton induction furnace (high frequency) at the Naval Research Laboratory and cast into a single ingot. In an effort to reduce the number of inclusions, and to improve and standardize the recovery values for the additions, each heat was given a "carbon boil" immediately after melt down. Also, as a possible aid to reducing the inclusions in the final material and to obtaining a finer grain size, a rare earth addition was made to the molten metal in the ladle prior to skimming and pouring into the mold. Each ingot was processed by forging to a slab having one dimension of the cross section four times that of the other dimension. After cropping top and bottom, 15 and 5 percent respectively, the slab was cut lengthwise and the central longitudinal section corresponding to one-fourth of the slab was discarded. The remaining two slab portions were hot rolled to oversize rods, annealed, straightened, and centerless ground to size. About 900 lb (408.6 kg) of finished rods were obtained for each standard as follows: 100 lb (45.4 kg) of rods 2 in. (5.56 mm) in diameter from the outer section near the bottom of the original ingot; 400 lb (181.6 kg) of rods 14 in. (31.75 mm) in diameter from the outer sections near the middle of the original ingot that is currently certified; and additionally, 400 lb of rods 14 in. in diameter which will be issued as renewal material when the first 400-lb lots are exhausted. As would be expected, the rod material 2 in. in diameter received far more severe working than the larger size of 14 in. in diameter, hence, the smaller rods exhibited a much finer grain size. The decision was made to confine the initial investigation by the electron probe microanalyzer and by the new quantitative metallographic techniques to NBS Spectrochemical Standard No. 461. The composition of this steel is shown in table 1. A random sample was chosen as typical of the entire 100 lb lot. Previous metallographic examination of several selected samples revealed no significant differences. It was desired to determine whether the steel was homogeneous in nickel and iron at micron levels of spatial resolution in the longitudinal, transverse, and normal directions. Furthermore, the average of a set of analyses for these two elements was to be checked against the certified value. For this purpose, the NBS electron probe microanalyzer was employed. This instrument, which enabled analyses to be made at the 2 to 4 μ level, has been described in detail elsewhere [2]. Since inclusions were found in the steel, they had to be identified; here both the electron probe microanalyzer and chemical-etching techniques were employed. Additional desirable information concerning the inclusions consisted of their volume percentage in the steel, the mean free path between them, and some idea of their number and size distribution. Concerning the steel itself, the following information was required: (a) a statement of the apparent ferrite and pearlite percentages in the worked structure of the steel, (b) the mean free path for ferrite and pearlite, and finally (c) the grain size of the steel. It is apparent that in order to obtain the required information, beyond the mere identification of the inclusions, one must resort to accurate quantitative metallography. 4. Experimental Procedure 4.1. Electron Probe Microanalysis To check the chemical homogeneity of the steel, both nickel and iron were investigated for uniformity of composition throughout the sample of 461. With the NBS electron probe microanalyzer, 46 separate spot determinations were made for nickel in the cross section and 62 determinations were made in the longitudinal section For iron, the corresponding determinations were 9 and 16, respectively. This gives a total of 108 separate nickel determinations and 25 separate iron determinations. -7 The electron probe microanalyzer was operated at 32.5 keV for all analyses. Specimen currents used were 3.5X10-8 (±1%) A for the cross section and 1.0X10 (+1%) A for the longitudinal section. Observed line to background ratios were 20/1 for pure iron and 30/1 for pure nickel using the second order unresolved Ka doublet for these elements in conjunction with an ADP crystal and a flow proportional counter. Using the first order unresolved Ka doublet for nickel in conjunction with a LiF crystal and a G-M counter, the observed line to background ratio was 210/1 for pure Ni. The probe size was 2 to 4 μ based on examination of the contamination spots formed. ADP refers to an ammonium dihydrogen phosphate crystal, which has a 2d spacing of 10.65 Å. The x-ray intensities from the sample were compared to those obtained for pure metal endmember standards. The corrections applied were as follows: Absorption: Philibert [3]. will cause the computer to demand a list of orders which are in the form of English words. Given this list, the computer stores the commands and carries them out in the listed sequence. To perform a command, the computer searches the order tape until it locates the matching title. It then reads Fluorescence: Birks [4] (enhancement of Fe-Ka in the detailed orders and executes this set of orders. by 1.73% Ni). Atomic Number [5]: Not required. Effect of continuous radiation [6]: Neglected. An appendix showing the detailed calculations for this paper is available on request [7]. 4.2. Computer Metallography A process which makes accurate quantitative metallography economically practical is one which employs a digital computer capable of accepting suitable photomicrographs and of printing out the desired information directly [8]. The computer which has been used to date is the SEAC (Standards Electronic Automatic Computer). Black and white pictures may be directly introduced in binary machine code by a scanning device. This equipment places a physical size limitation on the photomicrograph to be investigated since the scanning device can accommodate only one picture, 44 mm2, at a time. At present, 28 general operations may be performed on a suitable micrograph. Many of these have been described previously [9]. New operations used in this study are known as "SLICE," "BLOB TRANSFER," "BLOB CUSTER," and "STATISTICS" respectively and these will be described herein. The photomicrographs to be scanned must have the highest possible contrast between black and white areas because the accuracy of the machine is adversely affected by gray components. The size and shape of the areas to be shown as black must be truly representative of the sample at the magnification chosen. Furthermore, all grain boundaries revealed by etching must be shown as completed networks since the computer will find any openings and count such paired grains as single grains. The computer scans a raster of square spots 0.25 mm on an edge. For example, if the photomicrograph is taken at a magnification of 500 diam, the computer unit will be 0.25/500 or 5X10-4 mm which corresponds to 0.50 μ. Therefore, the magnifications used must be known accurately if data such as mean free paths are to be meaningful. To obtain the metallographic information desired, the machine was required to state the area of the black phase, to display the picture on an oscilloscope screen, to "slice" the picture horizontally and vertically, and to perform lineal analyses at right angles to each other. In certain cases, complete "blob" (any discrete black area) analyses were ordered from which complete data on each blob in the picture were deduced and printed out. To accomplish all of these orders, a master routine is fed into a portion of the computer memory. This The operator, in effect, causes the machine to execute literally thousands of orders in machine language by simply typing in a list of commands in English. To do the complete series of analyses required for obtaining all the information desired about one micrograph of the NBS-461 steel, the simplified English language command list to the computer is: SCAN. This causes the computer to be supplied with one picture 44 mm2. A blank border 1 mm wide is masked onto the picture. The net usable area within the border is 28,224 bits. AREA. This counts the number of black bits in the image. Division by 28,224 gives the percentage of the phase represented by the image. The area statement is checked by using a calibrated bar chart containing 14,112 black bits which is exactly half the usable picture area of 28,224 bits (704 SEAC words). DISPLAY. The image which is stored in the computer memory is exhibited on a cathode ray tube. DUMP. The image in the machine is electrically recorded on wire or tape. SLICE HORIZONTALLY or SLICE VERTICALLY. The slice operation separately determines the areas of 168 horizontal or vertical line. slices. The decimal print out classifies the 168 slices into a histogram of 27 classes. The upper limit of each class and the number of slices found therein is stated. The area is given in bits and in percent of the net image area (28,224 bits). A value, based on the histogram of the line slice distribution, is given in percent and the σ value of the mean area is given in bits and in percent. The o values may be expected to differ for horizontal and vertical slicings. The larger of the two a values is used. LINE. This causes the computer to treat the picture as an unbroken horizontal helical raster such as would result from wrapping the picture around a vertical cylinder and joining the right end of each scan line to the left end of the next line. This order produces a decimal table presenting a histogram (on √2 classes) of line lengths in the black and white portions of the picture. The mean free path is computed as well as the standard deviations of both path length distributions and of the mean free path. ROTATE. The picture is turned 90 deg as may be required for a directional interpretation. LINE. This second LINE order gives the lineal analysis at right angles to the first LINE order. BLOB TRANSFER AND RECORD. Each black blob in the picture is analyzed separately. The computer prints out the serial number, area, addresses of the corners of the circumscribed rectangle, height, width, estimated length (L), estimated thickness (T), and the shape factor (LIT). BLOB CUSTER. This operation reads back the data from the BLOB TRANSFER AND RECORD table together with an image of the single blob and computes the perimeter (P) of each blob as well as the complexity factor, P2/area. The table so constructed contains the input data for the statistical analysis. STATISTICS. The data from the BLOB CUSTER table are classified into histogram tables of eleven blob parameters on logarithmetically increasing classes of base √2. Automatic corrections for blobs cut by the picture edges are made. END. Machine stops. 4.3. Inclusion Counting The longitudinal and cross sections were polished for inclusion counting, i.e., to reveal the true number, size, and distribution of the inclusions. This was accomplished by performing the final polishing with a quarter-micron diamond powder on micro-cloth for long times at low speeds. The sample was alternately etched and polished and was finished in the unetched condition. One criterion for obtaining true photographic records of the surface was that two overlapping areas taken at different exposure times and printed for maximum contrast showed inclusions common to both areas as the same size in both prints. The micrographs used were below the limit where overexposure could conceivably blot out smaller inclusions and reduce the size of larger ones. Furthermore, when sharp black inclusions were obtained against white backgrounds by contact printing, the print truly represented the negative. Finally, the print was visually checked against the sample. When these requirements had been satisfied, the micrographs were ready for computer analysis (figs. la and 1b). Thirteen photomicrographs were processed. Six were of the longitudinal section and seven were of the cross section. Longitudinal section. Photomicrographs were prepared at 50X, 100X, and 200X. The 50X magnification encompassed the width of the rod in two 5X7 photomicrographs. The higher magnifications showed more of the detail of the inclusion shapes, however, the 200X pictures were too high in magnification to give truly representative values. Ac chosen for examination. Since the computer analyzes a central area on the picture of only 42X42 mm, the pictures were chosen to give maximum, average, and minimum blackness wherever possible. The six pictures chosen were: The criteria adopted for the inclusion polishing were as follows: (1) that there be no scratches in the area to be photographed when this area was sub-cordingly, four 50X and two 100X pictures were jected to rotation in polarized light, (2) that "open circles" were not true inclusions, (3) that the inclusions have sharp boundaries at the magnification of interest. When these criteria were satisfied, it was believed that a true representation of the inclusion concentration had been obtained. After being photographed in this condition, the samples were introduced into the NBS electron probe microanalyzer. 50X-maximum blackness 50 X-minimum blackness 50X-average blackness 100X-maximum blackness 50X-average blackness 100X-minimum blackness b FIGURE 1. NBS 461 steel polished for inclusion counting and microprobe examination. a. Cross section, unetched, X100 b. Longitudinal section, unetched, X100 |