Calibration Flashcards
What is quality assurance
Methods used to make sure the process is working and it up to a proper standard
What are the stages of data analysis when doing quality assurance
First you have raw data (your own measurements)
Then treated data (worked up data by finding concentrations and making a graph, might cut out some raw data)
Results (the final values and interpretations)
What are the steps to making a quality assurance prootocol
Set objectives (what are the goals, why are we analyzing, what do we need to know about the samples
Build specifications (what tests do you do, what min and max values do we want)
Make an assessment (stats, do the values agree with the specifications)
What do you include when trying tobuild specifications
You set a target value, min acceptable value, max acceptable value (tells the accuracy)
Max acceptable variance (tells the inconsistency in data)
What is a false postive
False negative
Test give “yes” value when actual sample should be no
Test give “no” value when actual sample should be yes
What is a blank sample
Sample with no analyte
Used to measure background signals/response
What is a reagent blank
A sample with all of the additional reagents (buffers, solvents) but still no analyte
Hasn’t been subjected to all preparation steps
What is a method blank
A reagent blank that’s also run through all of the processing steps (filtration, boiling, etc.)
Has everything except for analyte
What is a field blank
A method blank that is taken out into an environment that is similar to the analysis samples environment
What is a sample matrix
The rest of the sample (everything other than the analyte)
Can be simple like DI water or complex like ocean water
We need to account for the affects of the matrix components in blanks and the analysis solutions
What is a spike
- Identification: add more of the analyte to a sample and look for a change in response
- Spike recovery: adding known amount of analyte and seeing if the change in the signal is proportionate. If not, instrument might be faulty or the matrix could be in the way of the analysis
How do you makes sure you’re method of analysis is good?
Build a standard operating procedure and share it with other analysts
What is the error of selectivity
Give example
If the method of analysis responds to things other than the analyte
Ex. Measuring ph: ph electrodes respond to high [Na]
If low na then it’s fine
How do instruments respond to increasing analyte concentration
Linear response.
In terms of r^2 what is a good fit and what is a bad fit
1 is good
0 is bad
What do we want to see in residual plots
Scattered and random distributions of residuals
No patterns
What are methods to use to know if the value we measured is “true” and not just reproducible and wrong
Test the sample against a standard reference material (that been pre analyzed by trusted labs like NIST)
Test it against analysis done by another (trusted) lab (by sending it to that lab)
Test it against another trusted method (use an old method, spike recovery)
Test it with internal standards, standard addition, and spike recovery (to see if matrix or sample prep affect your method)
What are the two thing included precision and what do they mean
Reproducibility: repeating your own data, similar trial to trial or day to day
Repeatability: the similarity of results between different analysts or labs (other people repeating your values)
What is range
The concentration interval where the linearity, accuracy, and precision are all acceptable
What is included in range
Dynamic range
Working range
Limit of linearity
What is the dynamic range
The concentration range where there’s a measurable response to the analyte, even if the response isn’t linear
What is the working range
Concentration range where it includes the limitations of the sample prep or lab capability (not just the limitations of instrument response or the theoretical limits)
Where these condition can give a proper response
What is the limit of linearity (LOL)
The upper bound when instrument response stops being linear
Linear range is the range where the calibration curve is linear
Starts to plateau
What is the definition of sensitivity
It’s describes the lower limit
How small of a concentration or amount of analyte that we can reliably measure
If the concentration of analyte changes how much change it there in the signal
What is included in sensitivity
Limit of detection
Limit of quantification
Reporting limit
What is limit of detection
The lowest amount of analyte we can reliably DETECT that is significantly diff from the blank (give an actual detectable value)
The concentration of anaylte that gives a signal that is 3 times the sd of the blank (3sd)
How do you calculate concentration limit
What about signal limit
3s/m
B+3s
M is slope of the calibration curve
B is y int of the curve/mean blank value
S is the standard error of the actual signals from the samples in the experiment (not the blanks)
What is the limit of quantitiation
Lowest amount of analyte we can reliably quantify (know the concentration of)
LOQ, the formulas is 10(sd)
=B+10sy
What is the reporting limit
The lowest concentration we can get, below this, the concentration is reported as zero
What is an example of reporting limit
The reporting limit on fats is 0.5 but in canola oil it’s below this so it’s reported as zero
What is robustness in experiments
The ability of the method (experiment) to withstand small variations in conditions and still give acceptable results
Does the experiment run the same if the temperature changes?
If you find the limit of quantification using B+10sy, what does you do to convert the units
This equation gives units of whatever y axis is in, if we want to convert to x value units. Divide the answer by the slope (m)
What are calibration curves
Taking the instrument signal and finding the amount of analyte in the sample
Ideally linear (signal increases as concentration increases)
They have a large range and good sensitivity
The y intercept is mostly but NOT ALWAYS zero
How are calibration curves made
Measure the signal from A increasing series of standards that have known concentrations, also measure the blank (no analyte) and measure the analyte with unknown concentrations .
These standards of known concentrations are called external standards
What are the best practices for calibration curves
Check the plot and residual plot for outliers
Use =LINEST() in excel (it gives the error and it’s easier to exclude outliers)
Choose standards with a concentration range that the unknown concentration falls within
Measure the standards in random order (not low to high or high to low) this makes the slope bigger than is actually is
What is the recommended minimum to making a calibration curve
Measure 5 diff standard solutions and 5 replicates of each
When finding concentration of unknown from calibration curve how do you first make the graph?
Subtract each values absorbance from the blanks then make the graph
To find the error for the calibration curve calculation what do you do?
Use LINEST and delta X equation to find error
What is an internal standard
A known amount of a compound that isn’t the analyte that’s added to the sample.
Ex. Using F (standard) instead of cl (analyte) in the sample
The same amount of it is added to each standard (anaylte in known concentration) and unknown flask
Why does an internal standard help
Is help smooth out the variance between each trial of the experiment
Corrects for random errors caused by run to run variation
What is on the x and y axis of calibration curves with internal standards
X= [analyte]/[internal standard] or just analyte
Y= analyte area/internal standard area
What is standard addition
When a constant amount of the unknown sample is added to the standards (unknown is in every sample)
Then add increasing amounts of standards to each flask (0,5,10,15,etc.)
This means that 0ml of standard DOES NOT mean 0 signal since it still has analyte
Why do we do standard addition
The the matrix of the sample is complex (can’t replicate it in the standards) is makes it easier to just put the unknown in the sample.
Higher Analyte concentration overall might just help as well
What is special about calibration curves for standard addition
The y intercept does not equal zero
Added volume of standard (ml) is the x axis
The x intercept gives the equivalent amount of standard signal to unknown signal
What is selectivity
The extent at which a analytical method can distinguish analyte from other species in the sample (avoiding interference)
How do you find concentration of something given the signal and blank mean
The samples signal-the blank mean= the slope x concentration
Solve for concentration
If something falls so thin the control limits but not the specification limits is it still acceptable?
No it must fall within specification limits