Chromatography

Chem 121
Fall 2003

Definition and Overview

Techniques based on the physical and/or chemical separation

of 2 or more substances in a mixture
Scope and Purview:
-Theoretical Basis
-Instrumentation
-Qualitative and Quantitative Methods
n

We will focus on Gas Chromatography (GC) but will also

make reference to High Pressure Liquid Chromatography
(HPLC)

Theory: Basis for Analytical Separations

Separations are effected due to differences in substances

affinities for a mobile phase and a stationary phase.
Mobile Phase - typically a liquid (LC) or a gas (GC)
Stationary Phase - typically solid or a liquid

Sample MixtureA
B
(A & B)

Mobile Phase

Stationary Phase
Solid Support

Types of chromatography

The Partition Coefficient (K)

n

Allows us to quantify the distribution of a

compound between the stationary and mobile
phases:

K = Cstationary/Cmobile
large K = more time spent in stationary phase
= more time spent on the column
increased elution time = larger K
Assume that K is constant for a compound under
chromatographic conditions (thermodynamic constant)
5

Sample Chromatogram
Peak Width at
Half Maximum
(FWHM)

Retention time

Base
Peak Width

(thermodynamics)
Unretained
species (K=0)

W1/2

(kinetics)

Retention Behavior

tR: not a consistent measure of a compounds relative

affinity for stationary and mobile phases

n

varies with flow rate, temperature, etc.

Define a new term: Capacity Factor (k)

k = (tR - tM)/tM = tR/tM = k

Adjusted retention
time
7

Capacity Factor (k)

n
n
n

Gives a normalized measure of retention

Easily calculated from chromatogram
Relates directly to the Partition Coefficient:

k = K(VS/VM) = nS/nM
A thermodynamic property.

Bandbroadening

Affects peak width (W)

Governed by kinetic processes

We need to consider the effects of:

-Diffusion
-Eddy Diffusion
-Molecular Diffusion

-Mass Transfer
-time it takes to partition between the
stationary and mobile phases

Eddy Diffusion

Applies only to a column packed with particles (solid support

onto which the stationary phase is adsorbed)
Each compound species travels a different path through the
particles:

How affected by:

Particle Size? Mobile Phase Flow Rate?

( as dp )

(rel. independent)
10

Molecular (Longitudinal) Diffusion

Consider the effect of just a solute in a column
filled with just mobile phase (no packing) :

Mobile
Phase

AA
AA
AA
A
AA

A
AAA
A
A
AAA

Initial

Final

How affected by:

Diffusion Coeff (DM)?

Flow Rate (u)?

( DM )

( u )

-ALSO: DM varies with mobile phase temp., MW, and viscosity

11

Mass Transfer (Stationary Phase)

n

The partitioning process takes time

n

so, the rate at which a compound partitions into and out

of the stationary phase will affect bandbroadening

A
A

-A faster partitioning process results in decreased

bandbroadening
12

More Mass Transfer

n

Stationary phase mass transfer rate varies with:

n

DS - stationary phase diffusion coefficient

-increased DS gives increased rate (decr. broadening)

K - partition coefficient
-increased K gives increased rate (decr. broadening)

df - stationary phase thickness

- increased df gives decreased rate (incr. broadening)
u - mobile phase flow rate
- increased u gives decreased rate (incr. broadening)

13

Still More Mass Transfer

n

Also must consider Mobile-Phase Mass Transfer:

-varies with only two properties:
n

DM - mobile phase diffusion coefficient

-increased DM gives increased rate (decr. broadening)

dp - packing particle size

-increased dp gives decreased rate (incr. broadening)

How can we quantify each of these band broadening

components?

14

Theoretical Plates
Concept derives from distillation theory:

The height of a theoretical plate is the length of column

in which the equivalent to a single equilibrium
separation is achieved.
So, column efficiency can be gauged by the height of a
theoretical plate (H)
And, the separation power of a column can be assessed by the
number of theoretical plates (N), where:

N = L/H
15

Calculating N
n

The number of theoretical plates on a column is easily

calculated from a chromatogram:

N = 16 (tR/W)2
-assumes that peaks are Gaussian
-specific to a particular compound on that column
For non-Gaussian peaks (fronted or tailed peaks):

N = 41.7(tR/W0.1)2
1.25 + B/A

Asymmetry Ratio
(ratio of base widths on
either side of maximum)
16

The van Deemter Equation

Relates column separation efficiency to

bandbroadening components as a function of
mobile phase linear flow velocity:
H = A + B/u + (CS + CM)u
Eddy Diffusion
Longitudinal (molecular)
Diffusion

Mass Transfer
(stationary and
mobile phase)

17

The van Deemter Plot

HHPLC ~ 10x smaller

than HGC
BUT: NGC > NHPLC

uopt (gives Hmin)

Overall

(LGC >> LHPLC)

Mass Transfer
Eddy Diffusion
Longitudinal Diffusion
18

Resolution

This is the most critical figure of merit for a separation . . .

How do we define resolution?

RS = (2Z)/(WA+WB)

Z = (tR)B - (tR)A

19

RS Values versus Separation

RS = 0.75
RS = 1.0
(4% overlap)

Completely
(baseline)
Resolved

RS = 1.5
(0.3% overlap)
20

RS as a Function of Column Properties

n

If we:
n
n
n

Assume that WA WB
Express RS equation in terms of tR
Substitute in k and N where appropriate

We obtain:

RS = (N1/2/4) (( - 1)/) (kB/(1 + kB)

= Selectivity Factor = KB/KA = kB/kA = (tR)B/(tR)A

21

N and tR: Relationships with RS

n

We can solve the RS equation for N:

N = 16 (RS)2 (/( - 1))2 ((kB + 1)/kB)2

Even more rearranging and substituting allows calculation
of the retention time:

(tR)B=16(RS)2 (/(-1))2 ((kB+1)3/(kB)2) (H/u)

22

What Do We Want?
n

The object is to:

n
n

Obtain the MAXIMUM RESOLUTION

In the MINIMUM TIME

-alas, resolution and time typically work against

each other
n

Lets look at how N, k and affect these two

separation qualities

23

Effect of Changing N
n

Best resolution is obtained with MAXIMUM number

of theoretical plates (RS N1/2)
How can we increase N?
n
n

Optimize mobile phase flow rate (u)

Increase the column length (L)
-BUT: both methods also increase the retention time

DECREASE the height of a theoretical plate (H)

-increases column efficiency
-doesnt sacrifice time (tR H)

24

Effect of Changing k

Resolution increases with increasing k

How can we increase k?

n

Recall that k is related to the thermodynamics of the

partitioning process

Change:
n
n
n

Temperature (GC)
Mobile phase composition (HPLC)
Stationary phase composition

BUT: retention time ALSO changes with increasing k

HOW?
25

Optimum k

How do RS and tR
vary as a function
of capacity factor
(k) ?
Largest increases in
resolution occur with
k<5
Largest increases in
retention time occur
with k>5

Keep k >1,
>1, but also k <10

26

The Selectivity Factor ()

Optimizing k does not ensure that two compounds will be

resolved (they could have identical k values)
Need to consider :

= 1

B
A

k
GC: vary temperature
HPLC: vary mobile
phase composition

Temperature

27

Ocean Optics spectrometer

28

The General Elution Problem

n

How does one resolve ALL compounds in a mixture in which

there is a wide range of k values?

29

The General Elution Solution!

For GC: dynamically

vary temperature as
separation progresses

Temperature
Programming

45 oC

145 oC

30 - 180 oC

30

The HPLC Solution

For HPLC:
dynamically vary
mobile phase
composition as
separation
progresses

Gradient
Elution
31

Instrumentation
n

Fairly straightforward:

25 - 400 oC

32

Mobile Phase Supply & Delivery

For GC:

-usually use He or H2

-more efficient at high flow rates

than N2

-20 - 100 mL/min flow rates typical

-control with flow controller
-measure with soap-bubble meter

nFor HPLC:
-must be degassed and filtered
-0.1 - 10 mL/min flow rates typical
-PUMP: pulse-free, high-pressure (6 - 10,000 psi)
33

Sample Injection

For GC:

Syringe/Septum system

At temperature > column

temperature
Injection volume: 0.2 - 10 L
Nanoliter volumes for open
tubular columns

nFor HPLC:
Sampling loop/Injection valve
Injection volume: 5 - 500 L

34

Columns

Size
Packed (GC)
1 - 3 meters long
1 - 5 mm I.D.

Open Tubular (GC)

10 - 100 meters long
0.1 - 0. 3 mm I.D.

HPLC (packed)
100 - 300 mm long
4 - 10 mm I.D.

Support
Packed (GC)
Glass, Stainless Steel, Copper
<150 m diatomaceous earth

Open Tubular (GC)

Glass, fused silica

HPLC (packed)
Stainless Steel
3 - 10 m
35

More Columns!

Nonpolar

Polar

nStationary Phase
nHigh B.P. (stable at
column temps)
nNon-reactive
nSuitable K for analytes

Polar

Packed (GC)

4 - propanol

Coated onto support

1-10% of support mass

1 - heptane

Nonpolar

Open Tubular (GC)

Coated onto wall
36

Still More Columns . . .

n

Plates

Packed (GC)
~2,000 plates/meter
2 - 6,000 plates/column

Open Tubular (GC)

~2,000 plates/meter
20,000 - 100,000 plates/clmn

Sample Volume
Packed (GC)
microliters

Open Tubular (GC)

nanoliters

HPLC (packed)
microliters

HPLC (packed)
~50,000 plates/meter
5,000 - 15,000 plates/column
37

Detectors
n

Well consider three types:

n Thermal Conductivity Detector (TCD)
n Flame Ionization Detector (FID)
n Electron Capture Detector (ECD)
Classify with respect to:
n Selectivity
n Mass Flow versus Concentration based response
n Linear Dynamic Range
n Detection limit
n Destructive?

38

Thermal Conductivity Detector (TCD)

n

Measure change in thermal conductivity due to analyte gases

eluting from column (a differential measurement)
How?
-pass effluent over wire (tungsten or tungsten-rhenium alloy),
which is heated by a small current passing through it
-temperature of wire (measured as a change in resistance) changes
as the thermal conductivity of the effluent changes
-signal is based on a change in the temperature (resistance) of the
heated wires
-so, use a carrier gas with a VERY LARGE thermal
conductivity (relative to most compounds): He or H2

39

TCD Operating Characteristics

n
n
n
n
n

Simple
Universal (sensitive to almost ALL compounds)
LDR ~ 104
Non-Destructive
Detectability: ~10-9 grams (~10 ppm)
-only useable with packed columns
Concentration-Based Signal

40

Flame Ionization Detector (FID)

n

Ionize compounds in a flame

and measure current due to
ions produced
HOW?

H2/Air flame
Positive electrode to collect
electrons released by
ionization of analyte
41

FID Operating Characteristics

n
n
n
n
n
n

Simple
Selective (specific to organic compounds)
LDR ~106
Destructive (sample is burned)
Detectability: ~10-11 grams (~50 ppb)
Mass-Flow dependent signal

42

Electron Capture Detector (ECD)

Measure the
decrease in electron
flow from a source
due to the capture
of electrons by
eluting compounds
How?
63Ni

- + N2 2e- + N2+
(gives about 10-8 amps)
A + e- A- (decreases baseline current)

43

ECD Operating Characteristics

n
n

Complex (needs radioactive source, etc.)

Selective (specific towards compounds with
electronegative functional groups)
n

n
n
n
n

- e.g., organometallics, halide-containing compounds

(pesticides)

LDR ~ 102 - 104

Non-Destructive
Detectability: ~10-9 - 10-12 grams
Concentration-dependent signal

44

Other Detectors

For Gas Chromatography

n Mass Spec (well cover later)
n Infrared Absorption Spec
n Plasma Emission
n Etc.
For HPLC
n Refractive Index (universal)
n UV/Vis Absorption Spec
n Fluorescence Spec
n Conductivity
n Mass Spec
n Etc.
45

Applications
n

Qualitative Analysis
-retention times (tR) provide some qualitative info regarding
species identity:
n

Match tR values of unknowns with values from standards

run under identical conditions
BUT: matching tR values do not necessarily indicate
identical compounds
-tR values should match when run under different
experimental conditions (i.e., change temperature)

46

Example

47

More Qualitative Analysis

n
n

Biggest limitation: detector selectivity

Solution: increase dimensionality of detection
n Example: dual detectors

48

Kovats Index System

n
n

Provides a quantitative measure of retention

Calibrate retention using a series of standard compounds:
e.g., homologous series of straight-chain hydrocarbons

Define Kovats
Retention Index (I):

Pentane

Log tR

I = 100 n
So, I=500 for Pentane

Use to quantify retention behavior

relative to a set of standards

# carbons
(n)
49

Quantitative Analysis

Use peak area as the signal that correlates best with

analyte concentration
DIRECT ANALYSIS
n Calibration Curve: plot peak area as a function of amount
(volume or concentration) of analyte (from standards)
n

Determine amount of analyte in unknown by relating its

peak area to the amount of analyte on the calibration
curve plot
PROBLEM: large signal variations (poor precision) due to
injection uncertainties

Variable injection rates

Volume uncertainty
Pre-volatilization

50

Solution: Internal Standard

Add a known/fixed amount of a compound (the internal

standard) to all samples and standards
Normalize all standard and sample peak area values to the
internal standard peak area values
IF the internal standard behaves like the analyte, then it
should experience the same conditions and exhibit the same
signal fluctuations
Internal standard should have similar retention behavior and
volatility as analyte

51

Example: Internal Standard

7
6

37% RSD

5
4

Sample

37% RSD

3
2
1
0

Int. Std.
0.04% RSD

Ratio
1

Replicate Number

52

Area Normalization

If all compounds elute and can be quantified:

Measure peak areas for all compounds
Correct peak areas based on relative detector response
factors
Due to cmpd specific detector response
Calibrate using standards with known concentrations

Calculate results:
% composition = Cmpd Pk Area 100
Cmpd Pk Areas
53