Quality Control in Metals and Minerals

Sep. 10 2021

Definitions and Guidelines for the Interpretation of Quality Control

At Bureau Veritas, our core product is analytical data. Over many years, we have invested heavily in proprietary software and staff development to ensure that you get the highest quality data. Bureau Veritas uses a detailed and comprehensive quality system to minimize errors and maximize the reliability of our analytical results. This system applies a tiered approach to the application of quality systems in our laboratories. These tiers are layered in the following manner:

  1. Documentation in the form of standard operating procedures, underlying ISO 17025 standard, as well as training programs.
  2. The use of instrument calibration standards. These solutions are analyzed before any other solutions to establish the factors required to convert raw instrument data into concentration values.
  3. QC validation solutions are analyzed with client samples to validate each run and to confirm that each analytical run has been performed correctly. These are typically inserted immediately before and immediately after client sample solutions.
  4. Reference materials, replicates and blanks are inserted into randomly assigned positions within each rack as generated by our proprietary LIMS system so that they are analyzed with the client solutions. Their purpose is to provide a final verification of the entire sample handling process. These samples are made up of the following categories:

    • Sample preparation blank
    • Sample preparation replicate
    • Analytical blank
    • Analytical replicate
    • Certified Reference Material (CRM)
    • Internal Reference Material (IRM)
  5. Data review and validation. This is the final layer that is made up of sophisticated proprietary software and professional personnel reviewing the data. The following steps are applied:

    Software validation. Proprietary software is used to review the data for specific problems and to perform a series of rational checks on the data. Data values are flagged and given specific colors: red for fail and amber for warning. Operators must take action on failures and log their actions.
    Rack level validation is performed by the instrument operator that analyzed the samples. At Bureau Veritas, this person is a Chemist
    or other person with substantial and equivalent experience. This can only occur when the data has passed the software validation.
    The operator reviews the rack QC and validates the rack of samples if all QC samples pass.
    Method level validation. This validation is performed by the senior department Chemist. This review examines all racks analyzed by a
    specific method. Its purpose is to identify any trends or unusual results that are not apparent when only looking at a single rack of data.
    Final Job validation. This is performed by a Certified Assayer or equivalent senior person. This person has access to all the data from multiple analytical methods to check and compare. This is the person that ultimately signs the final certificate.

All Bureau Veritas laboratories and sample preparation facilities fall under the Quality Management Scope helping to ensure the same practices and procedures are followed throughout the organization. As of January, 2021, with the full alignment of the ISO 9001 and 17025 standards, Bureau Veritas has decided to maintain only ISO 17025 accreditation for its minerals facilities. Sample preparation locations are monitored regularly for quality control and quality assurance practices and are listed on the site-specific scopes of accreditation. To ensure ongoing quality assurance, Bureau Veritas participates every year in the CANMET and Geostats round-robin proficiency tests. We are a recognized participant in the CALA Proficiency Testing Program.

This article provides a detailed description of our application of reference materials, replicates and blanks.

The Use of Analytical Blanks and Preparation Blanks

Two types of blanks are used in the sample analysis stream for drill and rock samples. The first is a preparation blank that is collected from the cleaning sand or rock used between each and every job to clean the crushing and pulverizing equipment prior to starting another client’s samples. It also separates different jobs from the same client that may have been separated due to large differences in composition or grade. This blank appears as the first sample in each job, with results reported in the QC section of the certificate under the heading Prep Wash. The analytical results from this blank are used to monitor contamination during the preparation process. The second blank is an analytical blank which is inserted during analysis to monitor reagent contamination and is reported in the QC section of the certificate as BLK.

If the Client chooses to insert blank material, the material must have been previously certified by a minimum of 4 ISO 9001-certified and/or ISO 17025-accredited laboratories. The nominal maximum value for acceptance will be up to 1% of the preceding sample up to a maximum of 15ppb (preceding sample of 1,500ppb). For preceding samples above this range, additional cleaning rock must be run through equipment prior to these samples and repeat analysis will be at the cost of the client. In some cases, higher rates of contamination can occur. This is typically due to mineral types that contain higher levels of water of hydration (clay minerals). Our operators are trained to recognize this and use cleaning sand between such samples. Since this additional cleaning step carries an added cost, we do our best to contact the client to confirm these actions.

The Use of Replicates

Bureau Veritas uses analytical and preparation replicates on drill samples to track reproducibility of the analytical and preparation processes. Data for both types of replicates is provided with each certificate at no charge. Replicate precision varies with concentration from 100% or greater error at or near the detection limit for the method, down to the method precision at concentrations greater than 10 times the detection limit.

If clients choose to submit blind replicates, note that replicates on drill samples may not meet the same reproducibility criteria as CRM’s/IRM’s because the drill samples may not be as homogeneous as an aggressively prepared and mixed standard.

The presence of native gold can also cause serious reproducibility problems. Where the presence of coarse gold is suspected, the parties should discuss more appropriate analytical and preparation techniques that can mitigate these problems.

The Use of Certified Standard Reference Materials (CRM)

Bureau Veritas uses CRMs whenever possible to track analytical accuracy and precision for each method. If a CRM is not available or is of such high cost that they are not practical, we will use internal reference materials (IRMs) that are either synthetically made or certified by performing round robin analyses by several ISO certified and/or accredited laboratories. If an IRM is used, we routinely validate their concentrations using CRMs when they are available.

For concentrations above 10 times the detection limit, expected geochemical exploration sample precision is 15% for methods such as AQ300 and MA300. Ore grade expected precision is 7% at levels greater than 10 times the detection limit for methods such as AQ370 and MA370. Exact precision is method, element and standard quality dependent, so acceptance criteria for individual standard and method combinations are determined on a minimum of 30 replicates measured during the course of routine analyses at a single laboratory. It should be noted that the expected precision for gold in methods such as FA130, FA330, FA430 and FA530 are difficult to predict due to the heterogeneous distribution of gold in many materials.

Client Field Replicates

Field replicate precision is a measure of the sampling process and natural variability within the sample media; field replicates are not suitable for determining analytical precision.

Client Use of Blind or Hidden Internal Standards

Bureau Veritas encourages and strongly recommends the use of blind client standards and we recognize that their use is an important component of project data evaluation and acceptance. It is our policy to reanalyze any sample batch that contains a failed customer standard, free of charge, under the following conditions:

  • The client supplies us with the certification documentation for the standard or proof of certification parameters such as, but not limited to; method of analysis, number of participating laboratories, range of data in the round robin.
  • Standards must come from a recognized and reputable reference material producer, that is preferably accredited or certified as such. CANMET, CDN Resource Laboratories, Ore Research, Rocklabs, WCM and Geostats are recognized reference material producers. Certification criteria/method of analysis should be considered before determining if a standard is applicable to a method.
  • The analytical result falls outside 3 standard deviations of a population of no less than 30 values determined using a single analytical method (good laboratory practice indicates that 1 value between 2 and 3 SD’s is acceptable, while 2 consecutive values will call for reanalysis). In the above description, Bureau Veritas refers to the standard deviation of values determined over the course of these minimum 30 routine analytical measurements at a single lab, and not the value quoted in the certification sheet for the standard. This definition includes error associated with both the analytical technique, as well as error in the certified value, and is therefore a robust measure of a CRM’s performance under a particular set of analytical conditions. In addition, individual standard values that fall outside 3 standard deviations but still lie within the certified error of the material will not be considered to have failed QC validation and costs for requested repeat analyses will be borne by the client.
  • The failed standard is brought to our attention within 90 days of the initial reporting of the analytical results. If the reanalysis of a batch or rack is requested by the client due to a Standard failure and the only analytical result that changes significantly is the result for the Standard, the client will be charged for the reanalysis of the rack or batch as this indicates heterogeneity of the Standard itself. In addition, if both samples AND standards are unchanged upon reanalysis, the client will bear the cost of said reanalysis.

Some additional considerations should be noted:

  • Variability of a standard material is additive to the analytical method error. Therefore, a poorly prepared standard will increase the total standard deviation realized.
  • Selection of an appropriate standard that is both mineralogically and compositionally similar to the samples it is to be analyzed with is of critical importance.

If the standard has a different matrix then it would not be unusual if the only sample failing the performance criteria is the standard itself.

If the standard has a concentration that is not in a useful concentration range, then unexpected results can occur. For instance, if the concentration of the standard is too high, the laboratory may consistently reanalyze this standard under the assumption that the result is highly anomalous and therefore requires another check. This will waste money and time.

Determination of Method Confidence Limits to be Used for Pass/Fail Criteria

When referring to the Standard Certificate, neither the 95% confidence interval nor the standard deviation quoted in the certificate should be used to calculate control limits or to fail a batch of samples. The 95% confidence interval (normally appearing on the front page of a certificate) is a measure of the certainty of the accuracy of the recommended value. It does not relate to the expected precision during routine use. In addition, it does not account for variations controlled by the limitations imposed by a particular digestion method.

The control limits used to determine the passing or failing of batch data should be calculated from the data that is generated by the laboratory itself (see section “Client use of Blind or Hidden Internal Standards” above for details). Each laboratory provides Standards analyzed with each batch, for this purpose.

Whenever possible, the client should discuss their quality program with the laboratory prior to the start of the project. In this way, any difference in interpretation may be discussed and agreed to in advance.