analytical concepts Flashcards
analyte
chemical substance being measured
assay
process of determining amount of analyte in sample
qualitative analysis
identification of elements/compounds/etc in sample
quantitative analysis
determination of quantity of analyte in sample
signal
measured quantity which correlates to the amount of sample. Ex: absorbance, acid-base indicators
visual detection of signal (examples, pros and cons)
Ex: colour change, formation/disappearance of solid, other volumetric analysis
Pros: simple, low-cost, no maintenance
Cons: subjective leading to poor accuracy and precision, not sensitive, large sample volume required, time-consuming
Electrical detection of signal (examples, pros, cons)
Ex: voltage, current, transducer (converts light/heat/pressure to electrical output)
Pros: objective, highly sensitive, fast and automated, small sample volume
Cons: costly, maintenance and repairs (eg. calibration)
Noise
Variation in measured quantity. Aka standard deviation, denoted σ(bkg)
Background
Approximate constant base-level signal. Denoted µ(bkg)
S/N. How to improve it?
signal-to-noise ratio. Indicates validity of signal as being actually caused by analyte.
Proportional to sqrt(n) (n = number of measurements). Can be improved by signal averaging.
Valid S/N is >3
Detection limit
Amount of analyte corresponding to
S >= µ(bkg) + 3σ(bkg).
Setting µ(bkg) = 0 gives S/N>3
Matrix
All sample components apart from analyte
Blank
Man-made “sample matrix”
Positive control
Sample containing known amount of analyte (helps prevent false negative results)
False negative
Assay indicates no analyte when it is actually present
Negative control
Sample containing no analyte (helps prevent false positive results)
False positive
Assay indicates analyte presence when it is actually not present
Interference
Chemical in matrix which causes systematic error.
How does interference affect measurements (4 ways)?
- Acts on analyte or reagent
- Source of large background signal
- Cause negative or positive bias
- Cause absolute or relative errors (affects accuracy)
Selectivity
The extent to which other substances interfere with analyte determination
Masking agent
Prevents components in matrix from interfering
Accuracy
Closeness of expected value to true value
Absolute error (formula+example)
E = xi - µ
Ex: signal always 10% above true value
Relative error (formula+example)
E = (xi - µ)/µ
- Greater effect when signal is small
Ex. Always measures 2 units below true value
Precision
Agreement among results. Can be expressed using standard deviation (s)
Replication
Expected to give the same result in the absence of error.
Samples are…
- From the same source
- Run using the same method
- Under the same conditions
Random error
AKA indeterminate error.
Introduces uncertainty/stdev. Symmetric about µ.
Can be treated with statistics.
* Problem in precision!!
Systematic error
AKA determinate error.
Can be absolute or relative.
Skewed results, xi always either higher or lower than µ.
* Problem in accuracy!!
Types of systematic error (3)
Instrument error
- Calibration can minimize it
Method error
- Chemistry doesn’t behave as expected, something overlooked
- Difficult to ID
Personal error
- Incorrect data recording
- Deviation from established method
Confidence interval
Likelihood of sample mean being accurate to true mean. Computed with t statistic
Case 1 t-test
Compare sample mean to known value (from a reference standard).
t(exp)>t(table) means significant difference.
Case 2 t-test
Compare results from replicate analyses of same sample.
t(exp)>t(table) means significant difference.
F-test should be done first to verify that the precision/variance of the trials is the same.
F-test
Compares precision of two methods.
F(exp)>F(table) means the difference in precision is significant
Case 3 t-test
Compare means of paired data
1. Two methods used to measure different samples from same source
2. Measurements before and after drug treatment
G-test
First test done in statistical analysis, rules out outliers.
Find G(exp) of a sus point.
If G(exp)>G(crit), point is rejected.
Least squares analysis
Used to fit linear regression line. Assumes only error in y data.
Assumptions for fitting linear calibration curves via least squares analysis (4)
- Relationship between signal and quantity is linear
- Residuals are the result of random error affecting y
- Error affecting y is normally distributed
- Errors in y are independent of the value of x
Sensitivity
Slope of calibration curve. Signal per unit analyte
Dynamic range
Concentration over which calibration curve is useful (no extrpolation)
Selectivity
Given two compounds (1 and 2), the selectivity of a method of analysis is =m1/m2
Standard addition
Add known, increasing concentrations of analyte to sample. Plot line of best fit, x-intercept gives analyte in original sample
Limitations of standard addition
- Precise results only when amount of standard added is comparable to analyte quantity
- Time consuming, need multiple samples
- Dilution error
Internal standard
- Measure signals for both analyte and substance behaving similarly to analyte (known values)
- Find F value
Ax/[X] = F(Ay/[Y]) - Measure unknown sample signal (Ax)
- Spike with known standard (Y), measure signal again (Ay)
- Solve for [X] using F value
Can make Ax/Ay vs [X]/[Y] plot to average veriability
