Certified Reference Materials, Blanks and Duplicates

Certified reference materials, blanks and duplicates are complementary quality-control tools. A reference material provides a basis for checking specified properties; a blank investigates background or contamination; duplicates examine variability across the stages repeated. Their value depends on suitable selection, correct placement and interpretation within the sampling and analytical chain.^[1]

A successful check is evidence within that scope, not proof that every stage has performed correctly. For example, a prepared reference powder introduced at analysis does not independently test field collection or the crushing of routine samples.

Certified Values and the Properties They Describe

A certified reference material, or CRM, is sufficiently homogeneous and stable for its intended use. Its specified property values are characterized by a metrologically valid procedure, and its certificate reports those values, associated uncertainties and metrological traceability. Certification applies to the specified properties, not automatically to every value printed in supporting literature. NIST distinguishes certified values from other supplied values that do not carry the same status.^[2]

The selected CRM should match the intended measurement closely enough to challenge relevant method behaviour. Concentration, mineral matrix, decomposition and reporting basis all matter. A material certified for total element content does not automatically provide the correct comparison value for a partial extraction. The Joint Research Centre identifies this distinction when discussing the suitability of reference materials.^[3]

Certificate instructions are part of the comparison. Minimum sample intake relates to the demonstrated homogeneity at the portion size used. Moisture correction, storage and certificate validity also need attention. A small test portion taken contrary to the stated minimum may no longer have the certified value's applicable homogeneity and uncertainty support.^[3]

A well-characterized working reference material can monitor consistency, but assigning an internal mean does not make it certified. Independence also matters: reusing the calibration material as a check does not independently validate the same measurement procedure. NIST explicitly distinguishes calibration and validation uses of a reference material.^[2]

Comparing Results with Reference Values

Measured results should be evaluated against a suitable reference value with attention to both measurement and reference-value uncertainty. JRC's application note explains comparison on that combined basis. Lack of a statistically significant difference does not prove zero bias. The uncertainties must correspond to the result being compared, including whether it is a single determination or a mean, and be expressed on compatible uncertainty bases. Large uncertainty can conceal a difference important to the project; statistical agreement alone is not proof of fitness for use.^[4][5]

The uncertainty on a CRM certificate is not automatically the expected scatter of individual routine results. A control chart needs an appropriate estimate of routine measurement variability. Criteria for detecting a shift and criteria for deciding whether that performance is fit for the project are related but distinct. Eurachem treats ongoing internal quality control as a demonstration that a validated method continues to perform adequately.^[5]

Results should be reviewed by CRM identity, batch and method rather than pooled across unlike materials. A sustained offset may indicate a systematic problem even if the overall campaign mean appears acceptable. An isolated discrepant result can instead involve identification, handling or measurement; it requires investigation rather than immediate attribution to one cause.

Blanks and Their Position in the Chain

A blank is selected for known absence or sufficiently low, characterized content of the analyte relevant to its purpose. Visually barren material is not necessarily blank for all elements. Eurachem distinguishes blanks by composition and use, including sampling-related, procedural, reagent and calibration blanks.^[6]

In a mineral programme, a characterized coarse blank introduced before mechanical preparation can investigate contamination during the preparation stages it experiences. A prepared powder introduced after crushing cannot test contamination in that crusher. Similarly, a reagent or analytical procedure blank examines its own analytical boundary, not the recovery of geological material in the field.^[7][6]

For sampling-related checks, a field blank undergoes specified field handling, while a transport blank remains closed during the sampling visit. They do not test identical exposure pathways. Their relevance depends on the actual medium and collection procedure; every possible blank type is not required for every mineral sampling programme.^[6]

Blank interpretation should consider preparation and analysis order, neighbouring concentrations, background and detection capability. A non-detect blank does not rule out contamination below its detection capability that could matter to low-grade samples. An elevated blank is not a correction factor to subtract indiscriminately from unknown samples; contamination may affect them unequally.^[6][5]

What a Duplicate Repeats

A duplicate's name is less informative than its point of creation. The protocol should specify the material, selection step and downstream work performed independently. Here, an assay pulp means pulverised laboratory powder, not a processing slurry. Eurachem's sampling uncertainty guidance explains how repeated sampling and analysis can investigate different contributions, while also recognizing shared biases that remain invisible to repeats.^[8]

The coarse-split example corresponds to the stage described in ALS's preparation quality-control note; actual laboratory terminology should be checked rather than assumed.^[7]

Core fractions require particular care. Two halves or quarters may contain different local mineral distributions, and their masses and extraction conventions may differ. Nearby field samples may also include real spatial variability. Their differences should not be judged as if they were repeated measurements of a homogeneous solution.

Designing a Complementary Programme

Control selection should cover the material types, relevant concentration ranges and constituents needed for the engineering decision. A CRM suited to one matrix does not remove the need to assess a contrasting matrix. A satisfactory blank cannot demonstrate adequate sample mass. Close duplicate agreement cannot demonstrate trueness when both determinations share a bias.^[1]

The insertion strategy should permit timely detection and investigation of problems without relying on a universal percentage. Relevant considerations include batch structure, material variability, method changes, consequences of error and the amount of retained material. CIM's Canadian guidance recommends documenting pass/fail criteria and responses, rather than presenting one criterion for every project.^[9]

Independent laboratory checks add useful evidence when the submitted fractions and methods are comparable. However, two laboratories analysing portions of the same biased pulp do not independently validate the original sampling. A second laboratory is also not automatically a reference truth; unresolved differences need investigation of both methods and their uncertainties.

From a Control Result to an Action

The programme should state who reviews controls, which samples may be affected and how unresolved results are held or qualified. Retesting should begin from a material stage capable of addressing the suspected error. Repeatedly measuring the same solution is not an adequate investigation of an uncertain coarse split.

Original results, repeat results, certificates, control identities and the reason for the final decision should remain connected. CIM's database guidance emphasizes documented methods and identifiable selected assays when alternatives exist.^[1] This record distinguishes demonstrated correction from a convenient change in the reported number and allows later engineering users to understand the scope of the evidence.