Bookmark and Share
Standardization News Search
Feature
The Development of In-House Reference Materials in the Specialty Alloy Industry

by Tom Dulski

Industry’s needs far outweigh the availability of reference materials. Tom Dulski of Carpenter Technology Corp. shows one way of creating an adequate number of reference materials through traditional “wet chemistry.”

Introduction

The routine analytical methods used by today’s metal industries are, for the most part, faster, more accurate and precise, and certainly achieve lower detection and quantitation limits than any methods that have been developed in the past. They are, however, also much more dependent upon reference materials to attain these extreme performance levels. As a case in point, in the production of iron-, nickel-, and cobalt-base specialty alloys of the sort that have ended up in aircraft engines and prosthetic implants, a chemical analysis may involve the determination of 25 elements from major amounts to sub-part-per-million levels. The expected level of uncertainty may range from 20 percent of the amount present in the ppm realm down to less than 0.1 percent at the 30 percent concentration level. As the molten metal is being refined, the production personnel expect all this to be provided at near real time speed.

X-ray fluorescence and atomic emission spectrometers make this sort of feat possible. But even with the invocation of fundamental parameters calibration, few knowledgeable analysts would deny a critical role to reliable matrix-matched reference materials, particularly in validating high accuracy work with complex alloys.

The obvious first choice in most laboratory settings is to purchase certified alloy reference materials to satisfy all calibration and validation needs. Certified compositions in solid form are available from national agencies (such as the National Institute of Standards and Technology) and from commercial suppliers around the world. The COMAR database, maintained by BAM (the Federal Institute for Materials Research and Testing) in Germany, lists over 10,000 materials, including many metals and alloys. Unfortunately, except for simple matrices and a few high-volume production alloys, the reference materials available for sale fall far short of meeting the current requirements for the specialty alloy industry, which may involve the analysis of 500 or more standard grades of widely varying composition. Moreover, there are numerous modifications to base compositions and entirely new experimental alloys constantly being developed. Reference material needs here clearly represent a very wide, but shallow, sea that no government or private commercial source could ever adequately supply.

Choices

What choices, then, are available to laboratories that must play this frenzied game that demands such extreme accuracy? The modern mind’s initial response is to search for a technological solution—a new, non-comparative methodology that will churn out those 25 numbers over five orders of magnitude in virtually any combination, and do it with no sample preparation, memory effects, or down-time. The search, no doubt, continues. But the near-term solution lies elsewhere.

An important point is that care needs to be exercised in using current comparative technology. Most analysts know that making a reference material from another reference material incorporates a large potential for uncontrolled systematic error. But those same analysts may look upon calibration with pure element solutions as a simple, definitive answer to any analytical problem. However, in the case of compositionally complex systems such as high temperature alloys, such strategies must be applied with extreme caution.

One measure of the complexity of the analytical problem is the number of required analytes in three, admittedly somewhat arbitrarily defined, categories: major, minor, and trace. For purposes of discussion we can regard major elements as those present at greater than 1 percent concentration, minor as between 0.01 and 1 percent, and trace as below 0.01 percent. In this segment of the metals industry we can identify many alloys with five to eight major constituents and 10 or more minor components. It is also frequently necessary to measure more than 10 trace elements.

While the use of pure element solutions and comparative methodology obviously can be useful for such alloy sample systems it is clearly not simple and certainly not definitive. With sample systems such as these, spectral and chemical effects abound in most instrumental methodologies. These effects are compounded by the influence of long- and short-term drift, particularly on the response of major constituents. And, while software has reached a high level of sophistication, no correction model is known that can seamlessly handle all compositional combinations.

In addition, the preparation of high accuracy calibration solutions from pure elements and compounds is not the trivial task that it appears. High purity metals may bear undetected surface oxide contamination, salts may be deliquescent, and oxides may be non-stoichiometric. Commercially prepared solutions vary widely in quality, and are subject to the same pipetting errors and chemical compatibility issues in mixtures. Also, sole reliance on pure element calibration makes any systematic error in the dissolution of the metal alloy sample completely transparent. And this last issue can be extremely relevant since unique procedures are often necessary to dissolve some superalloy compositions.

What remains, then, as an alternative approach? It is time- and labor- intensive classical “wet chemistry,” used to certify in-house prepared alloy reference materials. Classical gravimetry and titrimetry, supported and supplemented by instrumental techniques that are augmented by separation science, can be the basis for producing a large custom-tailored suite of reference materials to meet all requirements for the high-speed production of specialty alloys. This solution to the problem, of course, depends upon the presence of an infrastructure of manpower, knowledge, and facilities that today is no longer commonplace. Such an investment has a significant return in the enhanced confidence that accrues to all the high-speed instrumental results.

Carpenter Technology’s Program

The way reference materials are made at Carpenter Technology Corporation involves a degree of systematic planning, involving close cooperation between production and research. In most cases the prime driver for the production of a new reference material is the process control laboratory where calibration and validation needs are usually the most urgent. Once the proposal for a new reference material has been accepted the search begins for a suitable starting material. Because micro-structural morphology can influence X-ray fluorescence results, we take great pains to select materials in the desired thermomechanical condition—in some cases producing separate reference materials from the same heat, compositionally certified separately as cast, wrought, and HIPed and forged forms. Starting materials may include billet sections from in-process production heats, finished bars from inventory, R&D heats of up to 400 pounds [180 kg], or other sources.

First, any undesirable surface features, such as end-porosity, “pipe,” and adherent scale are cut off. The material is then cut up into small solids of a size suitable for instrumental analysis, maintaining a record of their spatial location in the original work piece. Homogeneity testing typically involves both X-ray fluorescence and optical emission spectrometry. ASTM E 826, Practice for Testing Homogeneity of Materials for Development of Reference Materials, and similar criteria are applied to the data.

For small lot projects intended for infrequent use with rarely encountered compositions, only 30 to 40 solid specimens may be involved. In this range, 100 percent of the lot is homogeneity tested. However, with larger lots intended for the analytical control of high volume compositions, nearly 600 solids may be produced, and homogeneity testing is based on a sampling plan designed to represent the original uncut solid form. In the case of 100 percent testing, deviant data may lead to the rejection of one or two solids. In the case of sampled testing, deviant data may lead to 100 percent testing. But, in general, homogeneity testing is a “pass/fail” procedure.

About 100 grams of full-cross section millings are obtained from four or more pieces of “passed” material. The milling operation is performed slowly and at settings designed to favor uniform chip size and to avoid excessive heating. The combined millings are degreased with low-residue methylene chloride, dried, and mixed, and then submitted for chemical testing.

A Modern Classical Approach

With the exception of hot extraction methods for carbon, sulfur, nitrogen, and oxygen, all the analytical techniques employed in the certification process require the preparation of an aqueous solution of the alloy. This may involve hot-plate or microwave digestion with acids, molten salt fusion, or any of several specialized approaches. The complex compositions encountered in this industry segment often require very careful planning of the dissolution to ensure complete solvation of the analyte without loss as a volatile or precipitated species, and without the introduction of any interfering substances.

Our laboratory utilizes about 60 analytical methods that can be described as classical “wet chemistry.” Of these, about half would be considered definitive in the sense that they do not rely upon reference materials or pure element solutions to obtain a result, namely gravimetric methods, and volumetric methods based on normalities. The remainder are spectrophotometric methods that rely on a calibration curve prepared from pure element solutions.

In addition, we have developed numerous hybrid techniques that use classical separations to reduce background and circumvent interference and sensitivity problems with instrumental approaches. Many of these are the same or similar separation schemes as those used for classical gravimetric methods—ion exchange, mercury cathode electrolysis, solvent extraction, and precipitation reactions.

In inductively coupled plasma (ICP) emission the mercury cathode can be used to great advantage, particularly for low levels of aluminum, titanium, zirconium, and the rare earths. Traces of tramp elements can even be accessed by flame atomic absorption and very low levels of alloying additions by X-ray fluorescence. Such approaches can serve as a valuable cross-check on modern direct techniques like electrothermal atomic absorption and inductively coupled plasma–mass spectrometry (ICP-MS). In all, about 150 distinct procedures have been used in the Carpenter reference materials program. Confidence in a certified value can be significantly enhanced when completely independent procedures, especially those based on different chemical or physical principles, achieve the same result. It must be borne in mind, however, that analytical approaches can vary widely in expected uncertainty and in the cost and effort of their application.

Analytes are determined with a three-replicate minimum, and always accompanied by a complement of certified reference materials. We also attempt to account for the total mass balance in the alloy by exhaustively measuring all major and minor components. Thus, for example, we routinely determine iron in iron-base alloys. Our goal is a total certified composition in the range of 99.9 to 100.1 percent.

Reference Materials and Certificates

Once the analytical work is complete, a certificate is drafted following the guidelines of ISO Guide 31, Reference materials—Contents of certificates and labels, and ASTM E 1831, Guide for Preparing Certificates for Reference Materials Relating to Chemical Composition of Metals, Ores, and Related Materials. Uncertainties are included and calculated from a sum of both the analytical and homogeneity test variances. Our certificates are unique, however, in that the appended wet chemical test report contains all replicate data from the test material and all the traceability data obtained from the certified reference materials that were analyzed at the same time and in the manner as the test material. Two classes of certificates are issued: one for complete compositional analysis and one for “element-specific” compositional analysis. In the latter case, one, or only a small array, of elements are certified. In addition to the paper copy, certificate data are recorded in a searchable database that is accessible from any PC in the company.

The solid disks or rectangular solids are electrolytically etched with the company logo, a grade designation, and an identifying code number. These, and the corresponding certificate, are distributed to the various laboratories within the company where they are put into routine use for calibration, validation, and verification of comparative instrumental methodology. The remaining solid specimens are stored in a large inventory cabinet and are supplied to the company laboratories as needed. In some cases these materials are made available for sale to outside parties.

The unused portion of the chips is also stored and sometimes proves useful in re-check work, and for the validation or verification of routine solution work or thermal extraction analysis.

The Carpenter In-House Reference Material Program has been used to ensure the accuracy of product composition for many decades. As appropriate new methodologies and protocols have evolved, they have been incorporated into the program, extending its value and usefulness. Today, the program provides both a direct audit trail to NIST Standard Reference Materials and a database that documents our analytical capabilities. //

Copyright 2001, ASTM

Tom Dulski has been an analytical chemist in the metals industry for 38 years and for the last 27 years has been employed by Carpenter Technology Corporation in Reading, Pa. From 1994 to 2000 he served as chairman of ASTM Committee E01 on Analytical Chemistry for Metals, Ores, and Related Materials.

---------------------------

This paper was presented on March 4, 2001, at the Pittsburgh Conference on Analytical Chemistry (“PITTCON 2001”) in New Orleans, La.