Instrumental analysis

Main techniques
Spectroscopy
UV - VIS
fluorescence and phosphorescence
atomic spectroscopy (emission and absorption)
infrared
Raman
röntgen (X-ray)
radioanalysis
NMR
EPR
Main techniques

Electroanalysis Separation techniques

potentiometry chromatography
voltammetry (TLC, HPLC, GC, IC)
coulometry electrophoresis
electrogravimetry
conductometry

Other methods Hyphenated methods

Thermical analysis GC-MS
Mass spectrometry (MS) ICP-MS
Kinetic methods GC-IR
MS-MS
Solving an analytical problem

analyst ↔ customer

• definition of the problem

• selection of the method
• sampling, storage of the sample
• sample pretreatment
• analysis
• calibration (comparison with reference(s))
• calculation, statistical analysis
• providing the results, documentation
Analytical instruments

Main parts:
signal generator
input transducer
transformation modules
output transducers
Always electronic signal is measured: ( V or A)

The role of the computers in analytical chemistry:

collecting and processing the data
control of the measurement
designing the measurement, optimization
display of the data

it is more reliable and accurate than the human

Signal and noise

amplitude of signal

amplitude of noise

noise: random fluctuation of the signal

S average amplitude of the signal


N average amplitude of the noise
S  signal (V or A) 
(decibel)  10 log 
N  noise (V or A) 
Types of noise

eg. due to the random motion of electrons in the

circuit

eg. due to the drifting of the lamp in a

spectrophotometer.
It can be minimalized by beam chopping (the
chopper alternately sends light through the
sample and the reference cells.

eg. due to particles or bubbles in the capillary

or column in HPLC or CE
Noise reduction with signal averaging

Signal averaging can improve the signal-to-noise ratio

For the estimation of the noise level the maximum

amplitude of the noise in a region free of signal is
measured, and the signal is measured from the
middle of the baseline noise to the middle of the noisy
peak.

root-mean-square noise (standard deviation):

A
_ i is the measured signal for the ith data point
A is the mean signal
n the number of the data points

The averaging n spectra improves the signal-to-noise

ration by
Noise reduction with filters (hardwares)

Low-pass filter High-pass filter Band-pass filter

Experimental error
Types:
Random error: arises from the effects of uncontrolled variables in the measurements
Systematic (determinate) error: a consistent error that can be detected and corrected

xi  μ xi  μ
εi  xi  μ εi ,rel  %εi ,rel  100
μ μ
εi: error; xi: measured value; μ: true value

xi  x xi  x
εi  xi  x εi ,rel  %εi ,rel  100
x x
x : average of the measured value

assumption: μx

systematic error can be detected by: SRM, blank sample or comparison with other method
Experimental error

accuracy (nearness to the truth) and precision (reproducibility)

accurate and precise precise but not accurate

Gaussian curve
The lifetime of the electric light bulbs

0.09

mass distribution of hen's eggs

0.06
P

maverage = 58 g
0.03
standard deviation: 5 g ─
10 g ─
0.00
40 60 80
m (g)
Gaussian curve
z2
1  xμ
y e 2
z
σ 2π σ
y: the probability of a given value of x
σ: standard deviation
N N

 x i  x  
 i 
2 2
x  x
σ i1
σ i1
N N 1
if N ≤ 20

The m and s can not be measured, but

those can be approached by x and s.
The smaller the standard deviation, the
more closely the data are clustered about
the mean.
Discarding questionable data point
a) xi  x  4s
b) Q test
gap xi  xn
Q  xi: questionable value; xn: the nearest value
range xh  xl xh: the largest value; xl: the smallest value

N Qcrit
Confidence: 90 % 96 % 99 %
3 0.94 0.98 0.99
if Qcalc > Qcrit the
5 0.64 0.73 0.82 questionable point
should be discarded
7 0.51 0.59 0.68

9 0.44 0.51 0.60

Comparing replicate measurements (t-test)
Whether two sets of replicate measurements give the same or different results, within a
stated confidence level?

x1 , x2 : averages
n1,n2 : number of the data points
s pooled : standard deviation of the method using both sets of data
if tcalc > ttable , then the difference is significant (the 2 samples are different)
Quality assurance
Good Laboratory Practice (GLP), Good Manufacturig Practice (GMP)

• calibration, validation of the analytical instrument

• frequent calibration during the measurements
• use of proper standards, SRM
• analysis with different methods
• quality control samples
• accreditation

Quality assurance begins with sampling

Specifications: accuracy, precision, selectivity, sensitivity
Evaluation of the results

a) calibration curves (external calibration)

ideal case: linear, x=0; y=0 slope, intercept

standard solutions
blank
Evaluation of the results

b) single point standard addition

SxCs Cx: concentration of the analyte X

Cx  Sx: signal of the analyte X
Ss  Sx Cs: concentration of the added component
Ss: signal of the sample which includes the added component

c) standard addition

assumption:
the measured signal
should change linearly
with the concentration.

it is useful when
interferences from matrix
occur
Evaluation of the results
d) internal standard: known amount of a
compound, different from analyte that is
added to the sample
or calibration
e) isotope dilution:

AS: specific activity of the tracer

AX: specific activity of the sample obtained after the dilution
mS*: amount of the tracer
mX: amount of the analyte
specific activity: activity / mass