Chemical Reference Materials: Setting the Standards for Ocean Science (2002)

The preceding chapters detail the extensive need for reference materials in the ocean sciences. This chapter focuses on what is required to produce new reference materials as well as how to encourage the use of those that already exist.

Many of the reference materials currently available do not represent the matrices in which oceanographers work, nor do they contain the desired analytes at the concentrations found in nature. Furthermore, the sources of reference materials and certified reference materials are diverse and information about them is not well-coordinated. Better cooperation and linkages amongst reference materials users and certified reference material producers should be encouraged. COMAR (http://www.bam.de/cgi-bin/crm_office.cgi), an international database for sharing information between reference material suppliers is a good start but it is not “user-friendly,” and its existence is not widely known within the oceanographic community. More must be done to share information and encourage production and use of reference materials and certified reference materials. Scientific societies and international organizations have a clear and present opportunity and responsibility to encourage training in analytical quality control, and the provision and use of reference materials in ocean science, thus enhancing the quality of measurements made.

Reference materials must fulfill certain rigorous criteria before they are accepted and found useful by the analytical community. The following conditions are prerequisites for preparing reference materials that are mutually acceptable to organizations around the world:

1. Homogeneity. Homogeneity assures that the analysis of all subsamples of the reference material taken for measurement will produce the same analytical result within the stated measurement uncertainty. This is particularly important in the case of certified reference materials. Reference material producers therefore must specify the minimum amount of sample for which homogeneity has been measured and is valid. Finally, the ease of re-homogenizing the material after packaging must be taken into consideration.

2. Stability. Producers must state the length of the reference material’s useable life, since they can be sensitive to light, humidity, microbial activity, temperature, time, etc. Long-term testing is required to validate the stability of a material under a variety of storage and transport conditions.

3. Similarity to the real sample. To produce meaningful analytical results, the reference material should mimic as closely as possible the matrix of the test sample.

4. Accuracy, uncertainty, and traceability. A certified value is the best approximation of the true concentration of the analyte. During the certification process, a variety of analytical methods may be used to determine this true value. Uncertainty estimates ultimately based on this process, together with information about the material’s homogeneity can give a certified reference material traceability, needed for true international comparability.

Recently, several books have been written that describe the requirements and protocols for the production of reference materials in general and for environmental science in particular. Due to space limitations, only a brief summary of these ideas is presented below. The reader is encouraged to consult these works directly if a more detailed discussion is required.

Zschunke (2000) has collected a series of general interest papers on the use of reference materials in analytical chemistry. While written for chemists, the book addresses both the certification process and the application of reference materials to a variety of applications from materials testing to environmental analysis. A brief summary of international col-

laboration via the COMAR database is also included. For a discussion of the various steps of the certification process and the requirements for reference materials for a wide variety of specific substances see Stoeppler et al. (2001). Finally for a discussion of the preparation of matrix reference materials for various environmental applications, Parkany and Fajgelj (1999) provide a thorough discussion of example materials from several national metrology organizations along with chapters detailing their intended uses.

The preparation of a reference material requires substantial planning prior to undertaking a specific project (see Box 5.1). The process begins with the definition of the material to be produced, for example, “preparation of a seawater-based reference material containing the nutrient elements: NO₃, PO₄, and Si(OH)₄ at concentration levels appropriate to oceanic samples and certified for these constituents.” Such definitions arise either from internal decisions by reference material producers (such as NIST or NRC-Canada) typically in response to perceived needs, or through external pressure on these producers from potential users. (This report, for example, explicitly identifies a number of pressing needs for reference materials for the ocean sciences.)

At the same time, decisions must be made regarding the producers of these materials. For new reference material needs, identified within agencies such as NIST or NRC-Canada, this entails a decision as to whether to produce the material in-house or to sub-contract the work elsewhere. For externally recognized needs, both a supplier (e.g., either one of the recog-

nized national facilities or, possibly, an ad hoc group of investigators [see below]) and a funding source must be identified.

The next task is usually to obtain a sufficient amount of raw material with the desired properties. The amount of material needed depends on:

• the number of samples of reference material needed,

• the need for a feasibility study,

• the number of samples needed for the homogeneity study,

• the number of samples needed for the stability study, and

• the number of samples needed for characterization of the reference material (Stoeppler et al., 2001).

Once a sample is acquired, the supplier can begin to prepare the material in the state it will be used. It is particularly important at this stage to stabilize the material (if required) to prevent changes in composition of critical components and to homogenize the material so that all future sub-samples will be as identical as possible. If additional procedures (e.g., reduction of grain size) are needed to prepare the desired material, they should occur at this stage. At this point a preliminary assessment of homogenization should also be made.

Since one of the main goals of reference material production is to provide a stable reference material, tests for stability begin early in the production process. Ideally, these should be conducted over the expected lifetime of the reference material prior to its distribution; however, these tests can be conducted concurrently if required.

Once the stock material has been determined to be stable and homogeneous, it can be divided into portions appropriate for use as a laboratory reference material. The portion size depends on the analyst’s expectations. In some cases, providing single-use samples is appropriate: the material may have so little stability after being opened that it is unlikely that it could be used again. In such cases, however, insufficient quantity can lead to low detection levels, thus increasing error. Alternatively, however, a single package can produce several replicates so that too much material leads to extensive re-use of the opened container. This can result in problems with reference material integrity and thus cause potentially inaccurate results.

Once the material has been packaged, it must then be checked for uniformity. Any additional characterization of the material, in the form in which it will be distributed (e.g., grain size, color) should be conducted at this stage. Subsequently, the material is analyzed for the analyte(s) of interest for certification purposes, after which the certificate of analysis can be prepared. It is also imperative that a certified reference material or reference material be continually monitored for stability throughout its

useable lifetime. This process should also include periodic re-assessment of the certified concentration(s).

The useable lifetime of a reference material depends on both its inherent stability and also on its rate of use. It will thus be necessary to establish a timetable for the preparation of further batches of the material so as to ensure that supplies are always available. These subsequent batches (particularly if they are matrix-based) will need to be treated almost like a new material, i.e., they should be tested for stability, and a complete new certification will be needed for each replacement batch. This involves substantial work each time it is done, hence it is desirable to make as large a batch as practical, consistent with the probable shelf-life and expected usage.

There are several accepted methods for characterizing and producing reference values of reference materials and certified values of certified reference materials. The more widely accepted methods include:

1. Certification using one definitive method. This option is employed when a highly established, internationally accepted scientific primary method is available. The method must be shown to have negligible systematic errors and to provide sufficient measurement accuracy. An example of such a method is isotope dilution mass spectrometry.

2. Certification through interlaboratory testing. In this case, the reference/certified value is obtained by pooling results from several laboratories that have demonstrated capability in analyzing the analyte(s) of interest. The various laboratory means are manipulated statistically to determine the best or truest estimate of the value of interest.

3. Certification using at least two independent methods. At least two validated, robust, and independent methods are employed to produce the true value of the analyte.

Often, combinations of the above methods are used to certify reference materials. For example, the two-independent method approach is often coupled with the inter-laboratory testing procedure, after which the data are combined to give the proposed best value.

The rationale for choosing particular reference materials (and, in some cases, for recommending certification of reference materials) was discussed in Chapters 3 and 4. In this section, the committee proposes specifications for such reference materials, as well as some further suggestions specific to the preparation of the recommended reference materials.

Most of the reference materials discussed are based on natural matrices (seawater, algal cells, sediment) and would initially only be certified for a limited number of constituents. Nevertheless, it is apparent that such materials provide a resource for the investigation of a much wider variety of constituents, and it is important that the ocean science community be encouraged to investigate these materials further. In particular, the existence of these materials would facilitate a wide variety of necessary inter-laboratory method comparisons that have been neglected to date. Eventually these intercomparisons will result in consensus values for other constituents, which can then be assigned to the reference materials.

The preservation of nutrient solutions at the concentrations occurring in natural seawater is a major challenge to the routine production of a nutrient reference material. Preservation techniques must be developed that maintain concentrations stable for periods of at least one to two years. Gamma radiation will produce nitrite that is unstable. Therefore this method appears to be problematic. The feasibility of other techniques, such as autoclaving, ultra-violet or microwave radiation, freezing, and acidification, should be evaluated.

A single technique that preserves all three nutrients (NO₃, PO₄, and Si(OH)₄) in a single matrix would be most desirable. If this proves impossible, it will be necessary to use different approaches to preserve and store different nutrients. For example, glass containers cannot be used to store a reference material for Si(OH)₄ due to the slow dissolution of silica (Zhang et al., 1999), while the polymerization of silicic acid upon freezing eliminates that option for preserving stable Si(OH)₄ levels (Zhang and Ortner, 1998).

Nutrient calibration solutions in seawater are commonly prepared by dissolving known amounts of pre-dried, solid, primary standard salts in low-nutrient seawater. Low-nutrient seawater must be collected from oligotrophic open-ocean surface water to minimize background nutrient

contents. Such samples could be obtained from the Sargasso Sea during the summer after spring phytoplankton blooms have depleted most of the surface nutrients. Obtaining sufficiently low-nutrient seawater might require further biological depletion or chemical removal following collection. Quantification of trace nutrients in collected seawater (or at least demonstration that they are below the detection limits of most methods) will be necessary to obtain the high accuracy required for a certified reference material.

The only element whose concentration is recommended to be certified is iron. An iron reference material will also clearly be useful for studying other important metals such as zinc, manganese, copper, molybdenum, cobalt, vanadium, lead, aluminum, cadmium, and the rare earth elements. It is thus desirable to assure stability for some of these elements in addition to iron.

The optimal size for a portion of a trace metal reference material is 1L, so large volumes (greater than 1000 L) must be collected and prepared, which is feasible given currently available Teflon^® storage containers, sampling systems, and pumps. More experience is required, however, in collecting large volumes of uncontaminated water.

Preservation is a critical issue that has been carefully studied in the past. Acidifying materials with HCl to a pH of 2.3 (by adding 7.5 mmol acid to each liter of seawater) provides adequate preservation against precipitation or wall loss for most metals. Low density polyethylene bottles are currently considered to be acceptable for storage of acidified oceanic seawater. However, they require careful cleaning using hot surfactants to remove organic softeners and contaminants, as well as acids. They also require copious rinsing to remove contaminating metals. A pH of 2.3 is recommended for acidified seawater as a trade-off between stabilizing the sample and minimizing the aggressiveness of the stored seawater and the need for neutralizing the sample pH prior to analysis.

At present, isotope-dilution mass spectrometry provides the best method to certify iron concentration in the recommended deep-water reference material (an expected iron concentration of approximately 0.7 nM) and to obtain an information value for iron in the recommended surface water reference material (an expected iron concentration of approximately 50 pM or less).

Radionuclide reference materials can be prepared from natural ores of uranium and thorium. The uranium ore selected should have concordant ages, indicating that the system has been closed long enough to establish radioactive equilibrium through the long-lived nuclides. Specific measurements of uranium, thorium, protactinium, actinium, and radium isotopes are required to ensure that radioactive equilibrium is still present. The establishment of radioactive equilibrium for the thorium series takes only about 35 years, so thorium salts purified 35 years ago could be used for this solution. It is necessary to measure contamination from uranium series daughters in the thorium solution. For the ²¹⁰Pb / ²¹⁰Po reference material, the solution could be prepared from ²¹⁰Pb that is at least two years old. It is important that there be no ²²⁶Ra in this solution, otherwise the ²¹⁰Pb activity would not change in a predictable manner due to variable escape of ²²²Rn.

Each of the algal based reference materials recommended in Chapter 4 should be certified for both inorganic and organic carbon concentrations, total nitrogen concentration, δ¹³C of both the inorganic and the organic carbon components, and δ¹⁵N for the total nitrogen component.

These three biological matrices can be prepared by individually growing the T. pseudonana, S. trochoidea and E. huxleyi clones in large-volume photobioreactors, such as those available at NRC-Canada’s Institute for Marine Biosciences. Because growth conditions (light, nutrients, and temperature) can greatly affect the composition of phytoplankton, the photobioreactors must be tightly regulated to ensure that reproducible harvests of phytoplankton can be produced over time. Once harvested, the algal material can be freeze-dried and packaged for subsequent testing and distribution. Specific clones of cryogenically preserved microalgae should be obtained from a suitable source (e.g., Center for Culture of Marine Phytoplankton, Bigelow Laboratory for Ocean Sciences, W. Boothbay Harbor, ME) and used to inoculate the photobioreactors in order to minimize batch-to-batch variability caused by genetic drift in stock cultures.

Each of the sediment based reference materials recommended in Chapter 4 should be certified for both inorganic and organic carbon concentration, total nitrogen concentration, δ¹³C of both the inorganic and the

organic carbon components, and δ¹⁵N for the total nitrogen component. These matrix types (opal, carbonate, and aluminosilicate) provide end members for analysis and could be blended to produce any mixture desired. Sediment materials collected at the locations in Table 4.6 should be taken from within the top 10 cm of the sediment. Large box cores are the most appropriate for collecting a sufficiently large sample. The sedimentary reference materials could best be made available in a freeze-dried form, with known salt content, homogeneous at the mg level. To achieve homogeneity, sediments could first be ground and sieved through a plastic sieve to 42 µm, then mixed and either bottled or canned. Some unprocessed frozen sample could also be archived.

Although a detailed discussion of the economics of producing reference materials for the ocean sciences is beyond the purview of this committee, it was felt appropriate to indicate here the principal factors that influence the overall cost of a particular material.

As has been discussed, several steps are required to prepare reference materials. Each step in the production process carries associated costs that will vary according to the complexity required to produce the final material. Given the additional requirements of traceability and the rigorous certification criteria, a certified reference material can be much more costly to produce than simple reference and/or consensus materials. Although the exact cost of certified reference materials and reference materials are specific to each individual material, particular issues common to all materials should be considered.

The cost of developing and producing certified reference materials, especially matrix materials, is high. Researchers should realize, however, that in most cases the true cost of a certified reference material is never fully charged to the customer. For example, the costs of the research required to demonstrate the feasibility of producing a certified reference material are usually not incorporated into the final cost. As a result, a commercial organization may never want to undertake this initial research, since profits would often not be realized.

The cost of a certified reference material is influenced not only by the cost of its intensive certification procedure but also by collection, preparation, distribution, storage, stability assessment, the required annual or semi-annual reassessment, and general production costs. Organizations offering after-sales technical support bear additional personnel costs. To keep costs down, organizations producing certified reference materials have been trying to produce larger batches so that the various costs are amortized over a

larger number of samples. However, stability concerns may not always make larger batches a feasible option, particularly if the number of portions produced would not be distributed in an appropriate time frame.

Although the scientific returns made possible by the existence of wellcharacterized reference materials are clear, it is also apparent that the traditional approach in which reference materials are developed, prepared, and distributed by national standards organizations (such as NIST) provides limited value to the ocean sciences. Such organizations have many demands on their resources, and typically will not be able to place a high priority on the development, production, and distribution of reference materials for the ocean science research community in particular.

An alternate approach, discussed at length by the committee, directly involves the oceanographic research community in the process of development, production, and distribution of reference materials—ideally in partnerships with experienced reference material producers such as NIST or NRC-Canada. Appropriate science funding agencies could fund initial development costs, in partnership with various government and private “standards” producers. For example, in response to the needs of the National Status and Trends Program, NOAA contributed funds to the production of eight NIST Standard Reference Materials and seven calibration solutions. The SRMs were based on natural matrices and are prepared at two concentration levels. The calibration solutions are for each of the three chemical classes of analytes quantified by the National Status and Trends Program. These solutions were prepared at the request of the National Status and Trends contract laboratories and are currently available for purchase from NIST. In addition, the National Status and Trends Program also contracts with NRC-Canada to prepare unknown samples suitable for use in laboratory intercomparison exercises.

Direct participation of the oceanographic research community will enable collection of matrix-based materials such as natural seawater or sediments in an efficient manner (for example, material collection could be combined with other scheduled research activities). Furthermore, the extensive analytical skills of the oceanographic research community can be harnessed to conduct the necessary testing for homogeneity, stability, and ultimately characterization of the proposed reference materials. At the same time, complete success requires the participation of reference material producers, who bring to the table extensive knowledge and experience in the preparation and stabilization of reference materials, and on occasion access to necessary facilities.

This collaborative approach to the development of consensus materials could thus result in the eventual certification of those materials, if required. Arrangements could possibly be developed with existing certified reference material or reference material producers to disseminate information and to distribute materials to the oceanographic research community.

The development and initial production of any new reference material would require initial funding. Proceeds from the sales of the material could then be used towards the production of replacement materials. The full cost of the production and distribution of the material might be recovered in time, particularly for widely-used materials for which no further development is needed; the research activities using these reference materials would then bear future costs directly.

Funding agencies must commit to encouraging and funding both the development and distribution of reference materials and certified reference materials. However, production and distribution of reference materials will not become self-supporting unless the ocean research community uses them and widely appreciates their value. At present, this is not the case; therefore, any plan to produce reference materials must also be designed to encourage their use.

The first step in such a plan is education. Training in the field of oceanography needs to place greater emphasis on analytical quality control, particularly with respect to accuracy to complement the current focus on precision. For those already established in the field (and perhaps entrenched in their ways), national and international meetings should incorporate training courses and workshops.

Participation in round-robin exercises offers a substantial impetus for improvements in analytical quality control within individual laboratories. During these events, participating laboratories individually analyze samples of a particular test material. Such exercises must be organized using materials and analytes relevant to the ocean sciences. Laboratories must be encouraged to participate, even if they are at an early stage in their experience with the relevant analytical techniques.

Those scientists using available reference materials should be encouraged to report such uses explicitly in the scientific literature. A recent article by Jenks and Stoeppler (2001) goes so far as to suggest that scientific publishers should provide explicit recommendations as to how and where in a paper the use of certified reference materials should be described. Proposal and journal article reviewers also need to be encouraged to question the analytical quality control (and ultimate value) of measurements made without the benefit of reference materials.