Assay QA/QC Management

Overview

Monitoring of the results obtained from assay laboratories is essential to make sure that the results receive are both precise and accurate. Best practice is to use control samples that make up at least 5% of the number of samples submitted, as well as 5% of duplicates.

Contamination

Monitoring of the level of contamination requires the use of blanks. Blanks should have as close to a zero concentration of the property of interest as possible. The blank should also be of such a nature that it is treated in the exact same manner as the normal samples. For this reason it is preferable that the blank is coarse so that it passes through the crushing and pulverising part of the sample preparation procedure, as this is often where most of the contamination occurs. Commercially prepared blanks are preferable as the exact chemical composition of the material is known, but silica sand, or chips of quartzite or sandstone may be used if appropriate.

Blanks should return values at or very close to the detection limit of the assay method. Best practice is to insert a 5% of blanks into the routine samples, which results in a frequency of 1:20.

Accuracy Monitoring

Accuracy monitoring requires the use of standards, also known as certified reference material or CRMs. The standards chosen should have a concentration of the element or property of interest, similar to that expected in the samples. It is preferable to use a number of standards with a range of values that bracket the range of values you are expecting in your samples. Commercially prepared standards are supplied with the accepted value, confidence limits and an acceptable failure rate.

The results obtained for standards should be monitored to make sure they return acceptable values.

Best practice is to insert a 5% of standards into the routine samples, which results in a frequency of 1:20.

Precision Monitoring

Precision Monitoring requires the use of duplicates. Duplicates are used to make sure that the assay is repeatable, i.e. will you get the same (or similar) value twice if you re-assay another aliquot of the same sample. Since the re-assay is not the same sample as the original assay, you will not get the same result, but it needs to be sufficiently similar to the original result to give confidence in the repeatablity of the sample.

Consistently poor repeatability can point to too small as sample being taken, insuffcient crushing before splitting, or insufficient pulverising before splitting, contamination or inaccurate assay equipment.

NB Importing Assay Results

Laboratory assay results must be imported using Visual Geobase's Import Assay Results function in order to correctly match the assay results to the samples,standards, blanks etc. Please read this topic for more info : Importing Assay Results

Assay QA/QC Management using Visual Geobase

Standards and Blanks

The samples submitted as standards and blanks should be captured into the Submitted Controls table.

QC1.jpg

The sample number of the control, and the control name should be captured, as well as the batch number and hole_id if available.

QC2.jpg

The accepted value for the standards and blanks should be captured into the Standards - Accepted Values table.

QC3.jpg

The accepted value and confidence limits should be specified by the manufacturer of the standard. Sometimes + or - 10% is used if confidence limits are not known.

NB: The Measurement should be the same as the field name in the Sampling table for the program to correctly match the accepted value of the standard to the right assay result column.

QC4.jpg

Creating Control Graphs

Right click on the measurement of interest, and select either the datasheet to see the results directly, or one of the graph types:

QC6.gif

Control Graph

QC7.gif

Cumulative Deviation From Accepted Value

This graph type sums the difference between the result obtained and the accepted value. In the case below the standard generally over reports.

QC8.gif

Cumulative Deviation From Assay Mean Value

This graph type sums the difference between the result obtained and the average value obtained.

QC9.gif

Blanks

The graph below is an example of a control graph for a blank.

QC10.gif

Duplicates

The re-submission of an sample should be done under a new sample number, so that the analytical laboratory doesn't know what the result of the first assay was. The duplicates are entered into the SubmittedDuplicates table.

QC11.gif

The duplicate type, the original sample number and the new sample number of the duplicate are entered into the table. The batch number and the date submitted may also be captured if available.

QC12.gif

Creating Control Graphs

Right click on the measurement of interest, and select either the datasheet to see the results directly, or one of the graph types:

QC13.gif

X-Y Graph

A parity line or a regression line may be added using the options menu.

QC14.gif

HARD Graph

HARD is the acronym for Half Absolute Relative Difference. It is generated by calculating |x1-x2|/(x1+x2) expressed as a percentage and sorted from smallest to largest. Pairs that are identical will give a value of 0% and values completely different (i.e 0 and 1) will give a value of 100%.

The is a very useful graph for assessing the precision of a set of duplicates. The range of values graphed may be limited to exclude very high or very low values. Very low values close to the detection limit can give artificially poor precision (0.01 and 0.02 give a HARD value of 50% for instance)

As a rule of thumb pulp duplicates should have 90% less than 10% difference, coarse split duplicates should have 80% less than 10% difference and field duplicates should have 70% less than 10% difference.

QC16.gif

Q-Q Graph

A Q-Q graph means a quantile-quantile graph. This is useful for comparing two distributions to see if they are the same. Its is created by sorting the two sets of paired results and plotting them against each other. In the example below, the duplicate results returned slightly higher results that the original samples in the higher end of the assay range.

QC17.gif
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License