Fichas
Sampling, calibration, quality control and validation part 2
Fichas del conjunto (35)
- 1Internal Standard Calculation Modes Two point analysisAnalyse the analyte and internal standard.
Determine the response factor.
Analyse the sample with the internal standard added to it at a known concentration.
Calculate concentration of unknown based on signal ratio. - 2Internal Standard Calculation Modes Multiple point (graphical) analysisPerform series of calibrations with analyte and internal standard mixtures.
Produce calibration graph based on internal standard and analyte signal ratio.
Determine concentration of unknown from properties of the calibration graph. - 3Element response curve for semi-quantitative analysis by ICP-MS using two point analysis
- 4Internal Standard Procedure multiple point graphical analysisPrepare a set of standard solutions for analyte (A) as with the external std calibration curve method, but add a constant amount of a second species (B) to each solution.
Prepare a plot of SA/SB versus [A]. - 5Example: Quantifying Pb by ICP Emission using Cu as an internal standard
- 6No inter-element interference
- 7No Internal Standard Correction
- 8Internal Standard Correction
- 9Standard Addition CalibrationMost convenient when a small number of samples are to be analyzed.
Useful when the analyte is present in a complicated matrix and no ideal blank is available. - 10Standard Addition ProcedureAdd one or more increments of a standard solution to a number of sample aliquots of the same size. Each mixture is then diluted to the same volume.
Prepare a plot of Analytical Signal versus:
-volume of standard solution added, or
-concentration of analyte added - 11Example: Fe in Drinking Water
- 12Standard Addition ProcedureThe x-intercept of the standard addition plot corresponds to the amount of analyte that must have been present in the sample (after accounting for dilution).
The standard addition method assumes:
-the curve is linear over the concentration range
-the y-intercept of a calibration curve would be 0 - 13The concentration of Fe in the sample = ?
- 14Validating resultsNever trust an analytical result unless there is supporting evidence to show that it is correct.
Test yourself and your protocol;
-Spiked samples
-Reference materials
-Independent evaluation via proficiency testing schemes
-Accreditation to ISO 17025 - 15Validating your data – spiked samplesSub-samples to which a known amount of the analyte has been added by you
sample result 5mg/L + spike of the pure analyte 10mg/L = Spiked sample result 15mg/L
If not the correct answer – why not….
When to add the spike of pure analyte?
-Before prep or before analysis?
Issues with spikes in that it can be challenging to match them to the matrix of the analyte in the sample - 16Validating your data - reference materialsReference Material:
-Material, sufficiently homogeneous and stable with respect to one or more specified properties, which has been established to be fit for its intended use in a measurement process.
Certified Reference Material:
-Reference material characterized by a metrologically valid procedure for one or more specified properties, accompanied by a certificate that provides the value of the specified property, its associated uncertainty, and a statement of metrological traceability. - 17Use of reference materialsMaterials with known analyte concentrations
-after prep and analysis you should get the certified value
-If not – why not?
How confident are you to report your results to a client?
Reference materials provide an excellent independent verification of the quality of your work - 18Proficiency Testing Schemes
- 19Proficiency testing schemesWhat is the correct answer?
-Organiser has tested the material before release and established the correct answer.
-Requires a detailed testing program before sample release
Organiser has manufactured the test sample so knows what the correct answer is.
-Good where analytes degrade easily or are not all normally present at quantifiable levels in a single sample.
Organiser compiles the results from participants, removes outlying answers and establishes a mean value for each analyte. Then compares all results to the mean value(s).
-Good where labs are testing a wide range of analytes, gives flexibility - 20Sample resultsAll analytical measurements are wrong – it is just a question of by how much.
-All measurements have an error associated with them. Illustrate by means of error bars (typically set at 1 or 2 x the standard deviation of measurements of replicate preparations) or:
Derive an estimate of uncertainty.
-Most reliably derived from repeated preps and repeated instrumental analysis both undertaken over 10 days; this is unrealistic for u/g projects so a simple test is the preparation of a selection of samples in triplicate - 21Systematic and random error
- 22Accuracy and precisionAccuracy
-The closeness of the result to the accepted value
Precision
-The repeatability of a result
Accuracy and precision are not the same thing….. don’t mix them up. - 23Calculating error
- 24Basic statisticsTaking an average of multiple readings
Deriving a mean value
-How many readings should I take? – between 4 & 10
Describing the range the readings
-Their standard deviation (s for a sample, often SD)
-66% of the readings will fall within ±1s of the mean
-95% of the readings will fall within ±2s of the mean
-Most easily determined using a calculator or instrument software; the formula is: - 25standard deviation
- 26Standard deviationOften error bars are derived from a determination of the standard deviation of the measurement.
If a sample solution is measured 10 times, an error bar of ± 2s should encompass 95% of the readings made but is that the sum of the errors?
No it is biased towards instrument errors – what about the sampling and prep errors?
10 samples prepared and analysed would be a better measure of s
Not a robust measurement of method uncertainty
Often error bars are derived from a determination of the standard deviation of the measurement.
If a sample solution is measured 10 times, an error bar of ± 2s should encompass 95% of the readings made but is that the sum of the errors?
No it is biased towards instrument errors – what about the sampling and prep errors?
- 10 samples prepared and analysed would be a better measure of s
Not a robust measurement of method uncertainty - 27Error bars and data confidence
- 28Method validationA route of establishing that a method is fit for purpose.
Verifying that the laboratory is competent to perform the method.
Establishing a robust measurement uncertainty for the method to demonstrate the degree of confidence that can be applied to results. - 29Method validationSelection of the correct equipment
Production of clear documentation- methods & procedures for staff.
On-going staff training and on-going auditing of performance.
Clear evidence of measurement traceability.
Demonstrably able to derive accurate results
A determination of measurement uncertainty. - 30Method validationEstablish what the measurement requirements are.
-The measurand (the property to be measured).
-The expected amount of the measurand in the sample.
-The level of measurement uncertainty.
The above will dictate the feasibility of the measurement using the available facilities. - 31Method Validation The process of validation willConfirm that the method measures what it is intended to measure and that it meets its pre-set performance criteria.
Optimise method parameters.
Confirm the process of calculating results
Determine performance parameters such as sensitivity, limit of determination, bias/recovery, precision and uncertainty. - 32Method ValidationStandard methods will have been validated but for older methods the degree of rigour may not meet modern standards.
Even for modern methods the laboratory still has to demonstrate their competence. - 33Method Validation
- 34Method UncertaintyIdentify sources of uncertainty – mass, temperature, cal standards, pipettes etc
Quantify each uncertainty by either
-Reference data
-Replicate measurements (usually 11 measurements on duplicate subsamples of samples covering the working range and an analysis of variance analysis (ANOVA)
-Ensure an uncertainty is not double counted
Convert to standard uncertainties (u) expressed as standard deviations
combine the standard uncertainties to derive an overall uncertainty.
Then multiply the result by 2 to give an expanded uncertainty (with a 95% confidence). - 35—
Información
- Descripción
- Fichas sobre Sampling, calibration, quality control and validation part 2, creado por megan trent el 23/09/2020.
- Creado por
- megan trent
- Visitas
- 1
- Valoración