4th Annual Summer Symposium-Standrrdn
Standard Sample Program of the National Bureau of Standards HARRY A. BRIGHT National Biireair
qf
Standards, Washington, D . C.
The National Bureau of Standards program on standard samples was initiated in 1905 and has grown steadily until today more than 500 different standard samples of metals, ores, ceramics, chemicals, and reference standards are available for distribution to analytical and research laboratories. During the past year approximately 22,000 standard samples (units) were sold in this country and abroad. A brief review of the various National Bureau of Standards standards and their uses is presented. The preparation, analysis, and certification involved in the establishment of standard samples of composition such as metals, ores, and chemicals are discussed.
W
HEN the National Bureau of Standards was established-in 1901, the change from rule of thumb to technical control of
manufacturing processes was already well under viay. The insertion of chemical requirements in specifications to cover thr sale of manufactured products was also increasing. These changes necessarily had a profound effect on the development of chemical analysis, which had, up to these timrs, dealt chiefly with fundamental researches and examinations of naturally occurring materials. At a steadily increasing rate during the past 50 year*, the analyst has been called upon to determine policy In purchasing raw material and in grading finished products, and often his figures are required a t various stages of a manufacturing process This has necessitated many changes and advancw in analytical chemistry, because the great complesity of many industrial products has increased the difficulty of securing accurate determinations. Furthermore, it is necessary in many cases to obtain results as quickly as possible. Hence, there is need for special aids by which these requirements can be accomplished with the minimum expenditure of time. A major contribution to these problems is the use of NBS standard samples. HISTORY
The Kational Bureau of Standards program on standard samples began s m n after the organization of the bureau in 1901. The first standards mere prepared in 1905, when the American Foundrymen’s Association turned over t o the Chemistry Division of the bureau its project on the standardization of four types of cast iron. The distribution of these irons was assumed by the bureau after they had been reanalyzed by the bureau chemists and by a number of industrial and commercial laboratories. Very soon thereafter it was apparent from the numerous inquiries received that the preparation of other materials for the steel industry was worth while and the issue of various types of steels was begun. As the analysis and distribution of these irons and steels proceeded under the direction of the late W. F. Hillehrand, it was evident that there vias nerd for more standards, not only of metals but also of ores, chemicals, cements, and numerous other materials. Thus a program that was developed initially t o aid analytical chemistry has been expanded to cover a wide range of scientific work.
samples are materials that have been :tnalyzed carefully or whose phpical proprtir. have been accurately determined a t this h r e a u and, in many cases, also in other laboratories. Some are rertified for chemical composition; others are certified with respcct to a specific physical constant or property, as, for example, melting point, viscosity, index of refraction, heat of combustion, or othrr properties that are important in scientific research and industrial control Thr various t p r s of standards issued by the bureau and the iiumhrr of differrnt kinds of standardq in each group are qhonn in Table I S e n standards are added continuous11 to mwt increasing requirements, and many of thc older samples have been renewed a number of times For example, Bessemer steel 8g is the eighth issue of this type of steel, approximately 13,000 units (totaling 4400 pounds) having been issued since this standard was made available in 1907 The ninth renewal of this samplr, 8h, consisting of approximately 1000 pounds or thrre thousand 150-gram units, is now in process Typical of standard samples producrd to meet urgent demandR of industry are the two tin ores. Kos. 137 and 138, issued in 1943 At that time, the sources for importation of tin had changed entirely nith the lose of the Pacific areas from n hich tin had been obtained before the war. Consequently, government authorities and industry were having difficulty in comparing the quality of available tin ores. The assistance of the bureau TTas invited and ultimately the bureau’s standard tin ores provided the necessary basis for the evaluation of these new sources of tin
Table I. Types of Standard Samples Issued by National Bureau of Standards and Number of Different Standards in Each Group Composition Standards Chemical 49 Steels 9 Irons 10 Ferroalloys 20 Sonferrous alloys 10 Ores 22 Ceramics Primary chemicals 1.4
Other Standarda Hydrocarbons Radioactive Paint pigments Oils, viscosity p H chemicals Melting point materials Reference standards
191
33
29 13 5
5
2 341
Spectrographic Steels AI alloys Tins Total
SCOPE
Currently more than 500 standard samples of metals, ores, chemicals, and other materials are available ( 1 ). These standard 1544
52
4 10 200
-
V O L U M E 23, NO. 11, N O V E M B E R 1 9 5 1 Also, early in 1943, the Petroleum Industry War Council requested the National Bureau of Standards to include hydrocarbons of high purity in its standard sample program. Work on the preparation of such material was begun and within a year about fifteen hydrocarbon standards had been completed.. However, the needs for additional new hydrocarbon standards were increasing so rapidly that a much greater rate of production was necessary. Accordingly, a cooperative program between the bureau and American Petroleum Institute was effected whereby the preparation of standard hydrocarbons could be materially increased. This program was continued until July 1950, and there are now available 191 different standard hydrocarbons, certified for purity. Some of these compounds are certified also for density, refractive index, and heat of combustion. This type of cooperation with industry continues to be an important factor in the success of the NBS standard sample project. Only with such cooperation has the bureau been able to issue the large number of standards now available. With metals and ores, for example, industry offers suggestions in the selection of materials to be standardized, provides starting material without cost, and, after the bureau has prepared a homogeneous product in suitable form, participates in the extensive analyses required h(,fore the standard can be issued with a certificate of composition. The principal uses of S B S standard samples of certified composition and reference standards are: checking methods of analysis and analytical techniques; standardizing volumetric solutions; investigating improved methods of analyses; establishing the accuracy of new analytical methods; and calibrating and standardizing spectrometers, spectrographs, calorimeters, pH meters, Geiger counters, pyrometers, polarimeters, and other instruments. These primary standards are also used to calibrate secondary standards prepared and used by industrial laboratories that control the manufacturing processes of our complex industry of today. Furthermore, they serve to maintain uniformity of finished products, so that disputes between buyer and seller as to quality of material can be avoided or settled satisfactorily. Some of the specific uses of these samples in industry and technology may be mentioned. Standard samples of steels and iron control the quality of the steel industry’s output; standard hydrocarbons control the complex mixtures that go into plastics, synthetic rubber, gasoline for automobiles, and fuels for aircraft; and standard pigments define the colors of paints. Calorimetric and saccharimetric values are more easily determined with the use of the pure chemical, sucrose; and the dextrose sample aids in the determination of reducing values of various sugars. Standard samples of cement enable cement producers to standardize their fine-mesh sieves and surface-area measurements. Samples of specific p H value have facilitated attainment of greater accuracy and uniformity in p H measurements. REFERENCE STANDARDS
In addition t o chemical and physical standards, the bureau distributes a considerable number of standards of reference. Gloss standards have been issued t o control the finish of paints, paper, and other materials where reflectance is critical. Thermalradiation standards in the form of electric incandescent lamps are used to calibrate thermopiles, and high-purity benzoic acid in sealed glass cells provides a fixed point for the calibration of precise thermometric instruments. Radon, gamma-ray, and beta-ray standards are used t o check instruments for detection of these radiations and t o further research in this field. Standard color panels for kitchen and bathroom accessories help manufacturers maintain colors suggested in the bureau’s commercial standards, and dyed paper samples find use in standardizing fading lamps for determining the light-fastness of textiles and other materials. In the fiscal year of 1950 approximately 22,000 standard samples and reference standards were issued to users in this
1545 country and abroad. Of this number some 19,OOO units were standards prepared by the Chemistry Division of the National Bureau of Standards. A partial breakdown of the latter group into the various types of standards sold is shown in Table 11. PREPARATION OF STANDARDS
The preparation of standard samples involves many factors. I n selecting the types and composition of the samples to be issued, the bureau is guided by the number and kind of requests received and by the advice of those engaged in the industries interested. Also, the probable demand for the standard is expected to justify the time and expense of preparation. The particles of samples of metal standards must be comparatively small and more or less uniform in size, because, as is generally well known, composition often varies with chip or grain size, particularly with cast iron and nonferrous alloys. For example, the fine particles of cast iron cuttings in which free graphite has been blown out before sieving contain less carbon than the coarse particles. The degree of this variation in millings of different particle size of cast iron and leaded bronze is shown in Table 111.
Table 11.
Sales of NBS Standard Samples Prepared by Chemistry Division
(For fiscal year 1950) Kinds Steels, irons, ferroalloys Primary chtmicals Ores, ceramics Spectrographic standards Hydrocarbons Nonferrous alloys p H standards Others (6 types)
No. of Knits 8,138 2,796 1,551 1,427 1,073 942 439 2,896 19,262
Table 111. Yariations in Composition of Fine and Coarse Particles of Same Sample Fineness (sieve S o . ) Total carbon, 7’
Sieve 20 20/40 40/60
Cast Irona 14/20 2 45 Leaded Bronze Copper, 70
El 2 (9 8 77 1 60/100 71.5 Through 100 66 1 After free graphite has been removed.
20/30 2 27
30/40 2 04
Lead,
70
11 9 13.5 16 3 22 6 28 4
PERMANENCE
It is highly desirable that standard samples of any kind should undergo no change whatever during storage. None of the bureau’s standard samples meet this requirement in an absolute sense, but practically all undergo so little change that the consequences are insignificant in their intended use. Hygroscopic materials that release their absorbed water without any ot,her change during suitable drying treatment offer no problems. The same is true of materials that gradually change during storage, provided the composition of the sample can be restored by ignition to the same basis as that a t the time of standardization. Analyses of such samples can be certified on an “ignited” rather than on a “dried” basis. Examples of these are the soda-lime glasses, which gradually react with moisture and carbon dioxide to form compounds that are stable a t the usual drying temperatures. I n this case the analyst must determine the loss on ignition a t 1000” C. and correct all determinations for the difference between the value found and the value for loss on ignition indicated on the certificate. Careful inspection is maintained to guard against any change that might endanger the integrity of a standard. Unexpected
ANALYTICAL CHEMISTRY
1546
difficultiesare sometimes encountered-for example, a law-carbon ferrovanadium of relatively high aluminum content failed through slow oxidation within a year and was withdrawn. I n another instance, B sample of argillsceous limestone had been stored in tin-lined containem for years without ill effect, hut a sample of dolomite attacked the lining in a few months. METAL STANDARDS
A few words concerning the preparation of standard steel samples may he of interest. The steel is obtained in the form of round bars ahout 5 inches in diameter and 3 feet in length, weighing about 200 pounds. Six bars are eenerallv used for one standard and obviously must he
Fiewe 2.
Serrated Tool Used in Preparation of NBS Standard Samples
ard eight castings (approximately 260 pounds) comprised the starting material. ANALYSIS
The preparation of standard nonferrous alloys presents many obstacles, 8 major difficulty being that of obtaining homogeneous castings. Leaded bronzes are east in hollow cylinders with a wall thickness of about 1 inch. Phosphor-bronze 63b was cast centrifugally in hallow cylinders approximately 7 inches in length, with a wall thickness of 1.75 inches and an outside diameter of 8.125 inches (weight about 45 pounds). For this stand-
Methods used at the bureau for the analysis of standard samples are selected or designed to give the best obtainable results. There me, of coume, certain practical considerations involved in the degree of acouracy sought or required. For example, if the needs of practically all users are met by a value of 0.025% phosphorus in a standard steel, the additional expense and time i n v o h d in an attempt t o add the third significant figure are hardly justified. When possible, several methods of analysis for a n element are employed a t the bureau, Parallel oontrols are run on synthetic solutions of known composition or on earlier Ftsndards. A relatively large number of determinations are made-for example, in the analysis of low-tungsten steel 155, eleven elements were determined involving over 150 determinations, split into replicate runs oftwo to five determinations per set. No specified methods of analyses %re designated when the samples are sent t o the cooperating analysts, nor has any definite schedule of number of deter-
Figure 1. Lathes Used in Preparing Chips for NBS Standard Samples of Iron *+--I
V O L U M E 23, NO. 11, N O V E M B E R 1 9 5 1
1547
Figure 3. Machines Used i n Mixing NBS Standard Samples
minations, sets, or analysts thus far been requested. The cooperators, however, have been asked in recent years t o submit all results arranged t o show the number of runs in each set and the number of sets, both keyed to the analysts who made thcdeterminations. Although no systematic survey of these data has been made, i t may a t some time prove an inviting hunting field for a statistical Nimrod. Incidentally, since the practice was initiated t o have all individual values included, there has been a marked improvement in the quality of the values reported by the cooperating laboratories. VALUES ON CERTIFICATE
The appraisement of the vdues submitted by the cooperators for inclusion on the certificate of analysis is based on what may be called chemical judgment. This process involves, among other factors, (1)experience as to the accuracy that might be expected with superior handling of suitable methods, (2) interpretation of the probable accuracy of the bureau’s results, (3) survey of the data from all laboratories with respect to agreement, deviations, and ranges, and (4) setting up the limitspreferred forstsndard samples. Those analysts whose values are outside these limits are asked t o check their analyses. Finally, when the average value submitted by each laboratory for a n element is within the desired limits, or as close thereto as practical, the average of these individual averages is indicated on the certificate as the value for the percentage of the particular element. This is the general practice but is subject to occasional exceptions. The over-all average value on the certificate of snslysis obviously does not represent absolute oertsinty far the number of decimal places indicated-that is, if the average value of all the analysts for copper in a steel is indicated &R 0.062% it is not implied that the 2 in the third decimal place is absolutely correct, although occasionally 6ome individuals attempt to make such interpretations and criticize the bureau for failing to establish such a millenium. No rigorous limits can be indicated as t o the difference of the accuracy of the final averages from the true content. As the value on the certificate is a n average of a number of determinations by a number of procedures, these determinations are in turn useless for a statistical treatment to set up limits for the true content, unless i t can be shown that there are no additional precautions which would exert any determinate effect. However, i t would seem consistent t o consider that the true content is within the limits of 3 sigma of the individual averages reported by the cooperators. The procedure of setting up so-called “reasonable brackets” for ultimately obtaining the average value of all the cooperators’ results eliminates any intemretstion of the data for dispersion. precision, or scatter values.
The precision of the average values shown in Table IV for standard steels 8e and 8f is of interest, first, in connection with the general question of the accumcy of averaged values on NBS ccrtifiebtes and, second, with the degree of homogeneity obtained in preparing large lots of steel standards. These two standards were processed in two separate 600-pound lots from steel bars originating from the Same heat ingot. Standard 8e WE analyeed by a group of analysts in 1937 and standard 8f by a different group in 1941. I n theory, st least, identical values should be obtained, though obviously this does not occur practically. IIowever, as shonn in Table IV, excellent precision was obtained for the average values on the ten elements determined. This, of course, means a very satisfactory degree of homogeneity in both lots. Failure by the user to cheek the average value an the certificate could hardly be ascribed to nonhamogeueity, a claim that occurs now and then.
Table IV. Average of Values Reported by Cooperating Analysts for NBS Standard Steels 8e and 8P C Mn P
8e 0.074
8f
88
8f
0.009 0,003 Ni 0.420 0.004 0.097 Cr 0.005 0.099 0.003 0.080 V 0.002 S 0.081 si 0 . n ~ 0.013 Mo 0.001 0.001 a Proceased &Q two sewrate 600-pound lots Jrom the ~ z m e heat. Standard 8e anslyred in 1937, and 81 in l9ll by a dtfferent group of analysts. ~~
0.013 0.421
CU
0.008 0.004
~~
To summarmr: linm U I S G U ~ W L I , ~iueu, au s w n n p t has been made t o present a general picture of the scope of the National Bureau of Standards standard sample activity, and t o discuss in some detail several of the factors involved in the preparation and certification of standard samples of composition. Consumer acceptance of NBS standard samples is becoming more and more positive without any particular promotional efforts. This is believed to show that NBS standards are an integral part of our e.xpanding industrial economy and to demonstrate the
(1) Natl.
Bur. Standards, Supplment to C
