General Steps in a Chemical Analysis Mocked soil sampling experiment
General Steps in a Chemical Analysis 1. Identify question to be addressed by analysis (rarely just a number). 2. Select analysis procedure (experience, literature, regulations, etc.) 3. Collect samples (usually several replicates of multiple samples) 4. From each sample, take replicate homogenous lab samples. 5. Lab sample preparation: • digestion or extraction • separation and/or preconcentration 6. Analysis. 7. Assemble results into a report. 8. Interpretation and conclusions (considering step 1). **Analysis of unknown is meaningless unless you have collected the same properly, taken measures to ensure the reliability of the analytical method, and communicated your results clearly and completely.
Mocked soil sampling experiment summary of concepts: effect of sampling size and replication qalitative analysis quantitative analysis sample pretreatment precision and accuracy detection modes detection limit errors in analytical measurement soil = mixture of smarties and reeces pieces • yellow/orange = soil matrix • brown = Fe • green = PCB → very toxic • blue = Pb • purple = As → arsenic (inorganic = toxic, depending on form organic = toxic) • pink = Hg • red = dioxin → most toxic of all → assume all components have an equal molecular weight and are detected with same sensitivity Types of Analyses • qualitative analysis: identify components from their colours • quantitative analysis: counts the various colours • semi-quantitative analysis: get a rough estimate of concentrations (in-between)
Sample pretreatment • mix • “blind” sampling (no biases) • clean up: ◦ more difficult for large samples ▪ watch for overload of clean-up method ▪ takes more time and is this more costly ◦ required for large sample size for accurate quantitative analysis ◦ not required for qualitative analysis Detection modes • speciation analysis ◦ Fe(II) = smarties ◦ Fe (III) = reeces pieces • • •
non-destructive detector: visual observation equivalent to spectrophotometry destructive detector: taste equivalent to atomic spectroscopy selective detector: only detects certain analytes
Effect of sampling size on analyte concentration •
arithmetic mean: ◦ where n=number of measurements
•
median = middle measurement (or mean of middle 2 if even number of data points) ◦ less affected by outliers in small data sets spread or range = xmax – xmin
• •
standard deviation about the mean ◦ indicates dispersion about the mean ◦ indicates precision
•
relative standard deviation or coeeficient of variation:
•
absolute error:
•
E= x −x t x −x t x100 % relative error: E r = xt accurate but imprecise
accurate and precise
Accuracy vs precision • accuracy: how close the result is to the true value ◦ affected by random or indeterminat errors • precision: how reproducible the result is ◦ systematic or determinate errors
inaccurate but precise
innacurate and imprecise
Sample size • too small ◦ poor accuracy and precision ◦ increase in detection limit • large enough sample size ◦ better accuracy and precision • in both cases: the mean is closer to the true value, replicates are preferable Random or Indeterminate Errors • natural limitation of instrumentation, operator, etc • sometimes positive, sometimes negative but both equally probable • cannot be eliminated • ex: electrical noise causing a fluctuation in signal Systematic or Determinate Errors • unidirectional and can be identified and corrected • may be constant ◦ ex: excess of titrant for colour change; the relative erro is much smaller for big samples • may be proportional to sample size ◦ ex: interfering contaminants; error depends on fraction of contamination ◦ independent of sample size To detect constant errors • vary the sample size ◦ constant error has decreasing effect as sample size increases ◦ proportional error is independent of sample size Identifying/avoiding systematic errors Sources of determinate errors
Method of detection/correction
1. Instrument errors Calibration (periodic) ex: pipets, burets. volumetic flasks... with volume (due to response changes from wear, corrosion, different from calibration mistreatment...) 2. Personal errors (involving personal judgment) ex: end-point of titration
- electronic instruments - self-discipline
3. Method errors due to non-ideal chemical or physical behaviour of reagents and reactions - slowness of some reactions - instability of some species - non-specificity of reagents
Analysis of standard samples Analysis by independent method Blank determinations useful for: - detecting contaminants from reagents, vessels employed - correct for excess titrant for color change
Real environmental sampling • for simulation: ◦ sampling without replacement ◦ static population • in reality: ◦ amount removed is trivial ◦ one sample is unlikely representative of the area investigated: several samples are required ◦ dynamic population (depends on weather, time of day, etc.): the result is at best an estimate of analyte concentration Sampling consideration • lot: total material • gross or bulk sample: representative of the lot • laboratory sample: with exact same composition as the bulk sample • test portions Getting a representative sample • soil = random heterogeneous material ◦ differences in composition: ▪ occur randomly, on a fine scale ◦ random sample can be collected • for a segregated heterogeneous material: ◦ large regions have obviously different composition ◦ a composite sample must be constructed • when preparing a composite sample, a chemist would: a. take several samples at random from pre- planed locations in the bulk sample and analyze each. b. take representative samples from various areas of the bulk samples and combine them c. analyze as much of the bulk sample as possible • in a random heterogeneous mateial, a. differences in composition occur randomly and on a fine scale. b. large regions have obviously different compositions. c. samples are collected by taking portions from the desired number of segments chosen at random. d. a and c
Mocked soil sampling experiment summary of concepts: effect of sampling size and replication qalitative analysis quantitative analysis sample pretreatment precision and accuracy detection modes detection limit errors in analytical measurement soil = mixture of smarties and reeces pieces • yellow/orange = soil matrix • brown = Fe • green = PCB → very toxic • blue = Pb • purple = As → arsenic (inorganic = toxic, depending on form organic = toxic) • pink = Hg • red = dioxin → most toxic of all → assume all components have an equal molecular weight and are detected with same sensitivity Types of Analyses • qualitative analysis: identify components from their colours • quantitative analysis: counts the various colours • semi-quantitative analysis: get a rough estimate of concentrations (in-between)
Sample pretreatment • mix • “blind” sampling (no biases) • clean up: ◦ more difficult for large samples ▪ watch for overload of clean-up method ▪ takes more time and is this more costly ◦ required for large sample size for accurate quantitative analysis ◦ not required for qualitative analysis Detection modes • speciation analysis ◦ Fe(II) = smarties ◦ Fe (III) = reeces pieces • • •
non-destructive detector: visual observation equivalent to spectrophotometry destructive detector: taste equivalent to atomic spectroscopy selective detector: only detects certain analytes
Effect of sampling size on analyte concentration •
arithmetic mean: ◦ where n=number of measurements
•
median = middle measurement (or mean of middle 2 if even number of data points) ◦ less affected by outliers in small data sets spread or range = xmax – xmin
• •
standard deviation about the mean ◦ indicates dispersion about the mean ◦ indicates precision
•
relative standard deviation or coeeficient of variation:
•
absolute error:
•
E= x −x t x −x t x100 % relative error: E r = xt accurate but imprecise
accurate and precise
Accuracy vs precision • accuracy: how close the result is to the true value ◦ affected by random or indeterminat errors • precision: how reproducible the result is ◦ systematic or determinate errors
inaccurate but precise
innacurate and imprecise
Sample size • too small ◦ poor accuracy and precision ◦ increase in detection limit • large enough sample size ◦ better accuracy and precision • in both cases: the mean is closer to the true value, replicates are preferable Random or Indeterminate Errors • natural limitation of instrumentation, operator, etc • sometimes positive, sometimes negative but both equally probable • cannot be eliminated • ex: electrical noise causing a fluctuation in signal Systematic or Determinate Errors • unidirectional and can be identified and corrected • may be constant ◦ ex: excess of titrant for colour change; the relative erro is much smaller for big samples • may be proportional to sample size ◦ ex: interfering contaminants; error depends on fraction of contamination ◦ independent of sample size To detect constant errors • vary the sample size ◦ constant error has decreasing effect as sample size increases ◦ proportional error is independent of sample size Identifying/avoiding systematic errors Sources of determinate errors
Method of detection/correction
1. Instrument errors Calibration (periodic) ex: pipets, burets. volumetic flasks... with volume (due to response changes from wear, corrosion, different from calibration mistreatment...) 2. Personal errors (involving personal judgment) ex: end-point of titration
- electronic instruments - self-discipline
3. Method errors due to non-ideal chemical or physical behaviour of reagents and reactions - slowness of some reactions - instability of some species - non-specificity of reagents
Analysis of standard samples Analysis by independent method Blank determinations useful for: - detecting contaminants from reagents, vessels employed - correct for excess titrant for color change
Real environmental sampling • for simulation: ◦ sampling without replacement ◦ static population • in reality: ◦ amount removed is trivial ◦ one sample is unlikely representative of the area investigated: several samples are required ◦ dynamic population (depends on weather, time of day, etc.): the result is at best an estimate of analyte concentration Sampling consideration • lot: total material • gross or bulk sample: representative of the lot • laboratory sample: with exact same composition as the bulk sample • test portions Getting a representative sample • soil = random heterogeneous material ◦ differences in composition: ▪ occur randomly, on a fine scale ◦ random sample can be collected • for a segregated heterogeneous material: ◦ large regions have obviously different composition ◦ a composite sample must be constructed • when preparing a composite sample, a chemist would: a. take several samples at random from pre- planed locations in the bulk sample and analyze each. b. take representative samples from various areas of the bulk samples and combine them c. analyze as much of the bulk sample as possible • in a random heterogeneous mateial, a. differences in composition occur randomly and on a fine scale. b. large regions have obviously different compositions. c. samples are collected by taking portions from the desired number of segments chosen at random. d. a and c
