Adlard E.R. (ed.) Chromatography in the Petroleum Industry

Подождите немного. Документ загружается.

352

Chapter

(1)

Reversed phase: where the phase

of spherical silica with a non-polar

hydrocarbon chemically bonded onto the surface

less commonly

sty-

rene-divinylbenzene beads. The mobile phase is polar and

most often a

mixture of methanol and water, acetonitrile and water or tetrahydrofuran

and water. One important variant of reversed-phase

HPLC

reversed

phase ion-pair chromatography

(RP-IPC)

where the analyte is ionizable

protonizable, and the mobile phase consists of a buffered aqueous mixture

containing a counter ion of opposite charge to the analyte.

(2)

Normal phase: where the column

packed with spherical silica

with

silica with a polar phase chemically bonded to it. Typical bonded phases

include amino

(-NH,)

and nitrile

(-CN)

already referred to (and which

may also be used

reversed phase mode) and phenyl, nitro or diol. Some

specific phases such as dinitroanilinopropyl are also finding considerable

use. The mobile phase is non-polar, typically heptane with or without the

addition of small amounts

more polar solvents such as methylene chlo-

ride

or ethyl acetate.

(3)

Ion

exchange: consisting

sulphonate

quaternary ammonium func-

tional groups chemically bonded onto silica or styrene/divinylbenzene

polymer particles. Weak cations

anions can be separated without the

use of buffer solutions as mobile phase, whereas strong cations

anions

will require them.

(4)

Size exclusion

gel permeation: where solutes are separated by virtue of

their size in solution. This technique has many petroleum applications

for

the determination of the molecular weight of polymeric lubricant addi-

tives but is not considered

detail in this chapter.

The range

columns currently available is, therefore, extremely wide, such

that the separation of hydrocarbons, functional groups, ionic compounds, poly-

mers and even enantiomers can be achieved. Column design has advanced from

“conventional” columns to include disposable cartridges, radially compressed

columns, metal-free columns made from polyetheretherketone

(PEEK),

and col-

umns with adjustable end fittings which recompress the packing if voids de-

velop, prolonging column lifetime.

Most stationary phases are also available in microbore columns with internal

diameters of

1-2

mm, which offer the advantages of reduced mobile phase con-

sumption and greater mass sensitivity. By contrast, preparative scale columns

packed with any of the aforementioned stationary phases (with the exception of

diol) are also available

“off the shelf’. These have internal diameters (i.d.)

mm and can be used to recover larger quantities of analytes, either for fur-

ther purification

for

identification.

A petroleum

HPLC

laboratory can serve a diverse group of needs including

the analysis of fuels, lubricants, additives, waste water and refinery process

HPLC

and column

liquid

chromatography

353

samples. As a consequence one could easily expect to find silica, C18, C8,

-NH,,

-CN, ion exchange and size exclusion stationary phases in routine use,

each in one or more of the column designs described above.

12.2.5

Detectors

All detectors, be they for HPLC or any other analytical technique, must be

precise, sensitive and stable.

addition, HPLC detectors should have a large

linear dynamic range, be insensitive to temperature and eluent composition, ex-

hibit low noise and drift, and be simple and easy to maintain. Since the early

days

HPLC, no single detector has been able to fulfil all these criteria as the

flame ionization detector (FID) has done

admirably for GC. Instead, a range

of detectors has evolved, based either on changes in the bulk properties of the

mobile phase, or upon a selective property of the analyte(s). The subject has

been frequently reviewed and descriptions of the main detector types can be

found in any general HPLC text

[5,6].

The reader is directed to Scott for a more

detailed and mathematical treatment

[

81.

The treatment here will be restricted to

a brief discussion of those detectors which have found application in the petro-

leum industry. Even with this proviso, the majority of detector types currently

available

are

still included.

12.2.6

Selective property detectors

12.2.6.

W-visible spectrophotometers

Ultraviolet detectors have been used since the early years of HPLC and re-

main the workhorse detector in the majority of laboratories today. Early exam-

ples were effectively converted spectrophotometers with the only modifications

being associated with inclusion of a

flow

cell rather than a cuvette holder. These

instruments were therefore based on prism diffraction or grating interferometry,

such that the specific

A,,

of interest could be selected in order to achieve maxi-

mum solute sensitivity. The quantitation principle is the Beer-Lambert Law

which states that the amount of

or visible light absorbed will be directly pro-

portional to the solute concentration. Sensitivity and limits of detection will

therefore vary from solute to solute as a function of the individual compound’s

extinction coefficient. In extreme cases, where no

or visible light chromo-

phore is present in the solute, no absorption will take place and such solutes will

not be detectable. This is the chief limitation of

detectors and it is especially

apparent in petroleum analysis because saturated hydrocarbons have no chromo-

phore. A second major limitation is that mobile phases which themselves absorb

References

pp.

3 72-3 74

354

Chapter

light effectively impose a wavelength cutoff on the system. Below this

wavelength the absorption

of the mobile phase itself is

strong that solutes

with

A,,

the same region cannot be detected.

The principle of detection within prism

grating instruments relies on the

transmitted light of the chosen wavelength being cast onto a photomultiplier.

Only one wavelength

observed that

often a compromise between sensitivity

and selectivity.

I2.2.6.2

Diode

array

detectors (DAD)

Diode array detectors (Fig.

12.2)

effectively allow a much broader wavelength

range to be acquired simultaneously, such that an entire

spectrum (more

typically

200400

nm) can be captured repeatedly throughout the analysis. These

detectors became commercially available

the early

1980s,

have rapidly estab-

lished themselves as reliable and sensitive, and have allowed ever more complex

detection strategies to be employed.

course the solute still needs a chromo-

phore and the mobile phase

cutoff must be observed, especially in gradient

elution.

The principle of detection within the DAD relies upon an array of photodi-

odes of typically

0.5

nm resolution, such that the transmitted light after the flow

cell is dispersed by a holographic polychromator and directed onto the linear

photodiode array. Thus, by recording the signal output from one photodiode, the

eluent is monitored at a single wavelength and by recording the output from all

Photodiode Array

gg/gg'

Ellipsoidal Mirror

Elliosoidal Mirror

Deuterium Lamp

Fig.

12.2.

Optical

system

diode

array

detector.

HPLC

and

column

liquid chromatography

355

the diodes, an entire spectrum is obtained. The major disadvantage of this system

is that light of all wavelengths is present in the sample cell simultaneously and as

such, fluorescent light may well be present at the wavelengths being monitored.

practice this is such an infrequent occurrence as to be of little consequence

but it must always be borne in mind, especially if the mobile phase or solutes can

be excited.

The

DAD

can be used as a simultaneous multiwavelength detector to maxi-

mize sensitivity to each solute in turn, or to record entire spectra in order to ex-

amine peak purity, or to produce three-dimensional maps of wavelength versus

absorbance versus time. All three modes offer the user more accurate quantita-

tion than would be possible with a single wavelength dispersive

spectropho-

tometer. For research use, the ability to record an entire

spectrum of all the

unknowns

in a sample can give an early indication of solute identity. Coupled

with retention behaviour, this can yield hypothetical structural information, or

solute functionality, or carbon number, depending on the

mode employed

(normal or reversed phase).

12.2.6.3 Fluorescence detectors

light can interact with some solutes by exciting delocalized electrons into

higher energy states above the normal ground state. When these electrons relax

back to the ground state, the solute will emit most of the absorbed energy as light

at a longer wavelength than that which excited it.

solutes where this decay is

instantaneous or where it ceases immediately upon removal of the incident light,

the solute is said to be fluorescent. It is possible to monitor the emission wave-

length and filter out the excitation wavelength altogether, and this produces very

high sensitivity, some

two

to three orders of magnitude greater than

absor-

bance and is therefore

highly desirable method of detection.

order to take

advantage of the phenomenon, non-fluorescent compounds may be derivatized

prior to analysis with a reagent to produce a fluorescent derivative.

petroleum

analysis, fluorescence detection is most useful when the solute itself is highly

conjugated and fluoresces naturally, as do many polynuclear aromatic hydrocar-

bons. Fluorescent light emerges from the sample at random angles and most in-

struments monitor the light emitted at right angles to the excitation beam. Some

solvents have the ability to “quench” fluorescence such that the process is effec-

tively suppressed.

particular very polar or aqueous mobile phases and buff-

ered or ionic eluents are not recommended due to this phenomenon.

12.2.6.4 Electrochemical detectors

Compounds which are electrically oxidizable or reducible can be detected

electrochemically.

coulometric detectors the solute is completely electrolysed,

whereas in amperometric detectors, the solute is only partially electrolysed.

References

pp.

372-374

356

Chapter

Amperometric operation is more suited to flowing systems and

the commonest

mode of electrochemical detection. In amperometric detectors, the solute con-

centration is directly proportional to the diffusion rate of the solute across

the boundary layer to the electrode surface. The electrode current

therefore

dependent not only on the solute concentration but also on its diffusion coeffi-

cient.

detailed treatment of electrochemical detectors can be found in

Scott

[8].

Electrochemical detectors rely on the mobile phase being electri-

cally conductive and the most direct method of assuring this is to use buffer

so-

lutions.

12.2.6.5 Flame ionization detector

The use of the

FID

HPLC

necessitates the removal of the mobile phase,

chiefly by selective evaporation. Much effort has been expended into making the

FID

compatible with

HPLC

in order to take advantage of its properties of sensi-

tivity, known response and linear dynamic range. A number of mechanical trans-

port

systems have been developed originating with James

1964

[9].

moving

wire was employed to carry the column effluent through a heated zone where the

mobile phase was evaporated off, and then to another zone heated to a higher

temperature in order to evaporate/pyrolyse the solutes and carry them into the

FID

in a stream

nitrogen.

The chief disadvantages of transport detectors all lie with the transport

mechanism itself. The wire, chain

disk has proved to be difficult to coat uni-

formly, different solvents evaporate at different rates, accumulation of remaining

traces of solute give memory effects. These factors all contribute to relatively

poor signal to noise ratios. Since only a small proportion of the solute is evapo-

rated and detected, the sensitivity and large linear range of the

are not util-

ized. In conclusion the compromises inherent

transport

FIDs

have meant that

this detector is not widely used and its early promise for

HPLC

use remains

largely unfulfilled.

12.2.6.6 Mass spectrometers

The interfacing of mass spectrometers to

instruments

(GC-MS)

is possible

because both techniques are readily compatible.

GC-MS

is now one of the most

powerful diagnostic tools available to analytical chemists.

Interfacing

HPLC

to mass spectrometers

(LC-MS)

is much more difficult and

has largely hinged on the design of liquid phase separation systems and removal

eluent until relatively recently. Work on

LC-MS

began in the late

1960s

but it

was

not until the work of Homing et al.

[lo],

Scott et al.

13,

and Arpino et al.

[

in the

1970s

that

LC-MS

was effectively achieved.

Overcoming the relative incompatibility

the liquid phase eluent and the

high vacuum required

the source

the MS has proved to be a severe chal-

HPLC

and column

liquid

chromatography

357

lenge. In addition, the higher molecular weight, lower volatility and chemical

polarity of many compounds separated by LC make them less easily ionized than

the compounds amenable to GC-MS. Because of this, electron impact ionization

(EI), which is

successful in GC-MS, has proved less

in LC-MS. More rele-

vant ionization techniques such as fast atom bombardment (FAB) and atmos-

pheric pressure chemical ionization (APCI) have been applied in order to sur-

mount this problem. The mere modification of

to make it more compatible

with existing high vacuum, electron impact MS has not on its

own

proved suffi-

cient, and the development of more compatible MS ionization and inlet systems

has been necessary for the

two

techniques to merge successfully. The whole

subject has been well reviewed recently by Niessen and van der Greef [13].

These authors list

distinct types of interfaces for LC-MS developed since

1972. The reader should consult reference

[

131 in respect of thermospray LC-MS

and the particle beam interface, both of which have been successful in petroleum

and coal-based applications.

12,2.6.7

Injured and

NMR

photometers have found little use as HPLC detectors for

two

main reasons.

Firstly, most solvents used as mobile phases absorb in the most useful regions of

the

spectrum. Secondly, using absorption wavelengths away from solvent ab-

sorption bands has invariably resulted in less sensitivity and higher background

levels. The exceptions to this have been where the analyte contains a carbonyl

(C=O) group and in size exclusion chromatography of polymers. The former

case is able

take advantage of the high extinction coefficient and hence high

sensitivity of the carbonyl group. The latter application is able to overcome both

low

sensitivity and high background by virtue of the relatively high sample con-

centration required by SEC.

Many of the limitations have been overcome or greatly reduced by Fourier

transform infrared (FTIR) instruments. Modern FTIR spectrometers have signal

to noise ratios over 100 times larger than energy dispersive instruments and as

consequence sensitivity is greatly improved. Their chief disadvantage is that of

high cost and another disadvantage is incompatibility with reversed phase elu-

ents. The combination of water absorption and band broadening due to hydrogen

bonding conspire to reduce sensitivity and to limit the usable part of the

spectrum.

Proton nuclear magnetic resonance spectroscopy

('H

NMR) has also been

used as an on-line HPLC detector. This technique exploits the odd spin of the

hydrogen nucleus,

lH,

in order to gain information on the environment of various

hydrogen atoms in the analyte molecules.

this way, the signals due to methyl,

methylene and aromatic protons in various molecular environments can be sepa-

rated and quantified. Once normalized, the proportion of various hydrogen types

References

pp.

72-3

358

Chapter

can be calculated and the alkyl, aryl and heteroatom substituents present in a

sample elucidated. Proton

NMR

will be unable to distinguish hydrogen atoms

from the mobile phase from those of the analyte and these will be included erro-

neously in the normalization if present. For this reason, static NMR experiments

or LC-NMR cannot use standard solvents but are required to use perhalogenated

or perdeuterated solvents. This

a severe limitation to on-line LC-NMR since

these solvents are extremely expensive, especially if significant volumes of per-

deuterated solvents such as chloroform-d (where

99.8%

of the hydrogen

re-

placed by deuterium) have to be used for the LC separation. Another consider-

able limitation is the high capital and running cost of a modern Fourier transform

NMR spectrometer. Nevertheless, this technique has found application in petro-

leum analysis and is expected to find increasing use.

12.2.7

Bulk

property detectors

12.2.7.1

Refvactive index detector

The refractive index detector remains the second most widely used LC detec-

tor after the

detector. It is universal, detecting all analytes whose refractive

index

(RI)

differs from that of the mobile phase.

The

of a substance

a dimensionless constant which typically decreases

with increasing temperature. Three types of

detector are available and all are

termed differential refractometers, that is they measure the difference in

RI be-

tween a sample cell and a reference cell containing mobile phase only. It fol-

lows, therefore, that all refractometers are sensitive to temperature changes and

changes

eluent composition. Thus

in order to use them for gradient elu-

tion, the reference cell must always contain a mobile phase of identical compo-

sition to that in the sample cell and this is often impossible to achieve. For good

baseline stability,

detectors are thermostatically controlled, either by a water

bath or by an insulated cabinet.

Deflection or angle

deviation instruments have a split flow cell, with Sam-

ple on one side, reference eluent on the other. Light from the source passes

through this cell to a mirror behind it and

reflected back through the cell to a

photomultiplier. If a solute of different

enters the sample cell, the light beam

will be deflected. The photomultiplier output is proportional to the magnitude

the deflection. Deflection

detectors are simple and have a wide linear dy-

namic range. Instruments manufactured by Waters Associates have typically

been of this design.

Fresnel refractometers pass parallel incident light through a prism onto sam-

ple and reference simultaneously. If the refractive index of the liquid in the

sample cell differs from that in the reference cell, some light from the sample

HPLC and column

liquid

chromatography

359

cell will be diffracted, reducing the intensity of the beam reflected back out of

the sample cell. The difference between the intensities of the sample and refer-

ence beams is measured by a photomultiplier and recorded. The linear concen-

tration range of this detector is less than that of the deflection instrument unless

two separate prisms are used to cover the entire

range. The optical cleanliness

of the system is also more critical than for the deflection detector. Fresnel refrac-

tometers have been manufactured by Perkin Elmer.

Interference refractometers split the source beam, pass it through sample and

reference cells simultaneously and then recombine it.

Any

difference in refrac-

tive index between sample and reference cells will manifest itself as a difference

in optical path length when measured by an interferometer. This design is more

sensitive than the previous types and additional sensitivity is possible if

laser is

used as the light source as by Woodruff and Yeung

[

14,151.

summary,

detectors are universal and can be sensitive under carefully

controlled conditions. Their use in gradient elution is still far from straight-

forward and base line drift is to be expected when the mobile phase composition

changes even by relatively small amounts. Despite all these operational draw-

backs, they

are

still the detector of choice when the solutes have no

chromo-

phore, especially in isocratic determinations of saturated hydrocarbons.

12.2.7.2

Evaporative

light

scattering detectors

The evaporative light scattering detector (Fig. 12.3), evaporative analyzer or

mass detector was developed and patented in 1966 by Ford and Kennard

[

16,171.

It was not until 1978, however, and the comprehensive work of Charlesworth

[18] that its usefulness as an

HPLC

detector was fully realized. The theory

operation, construction and performance of what is now referred to as the “mass

detector” can be found in that reference.

essence, this type of detector consists of a nebulizer, evaporation chamber,

light source, scattering chamber and light trap and a photomultiplier set at 135”

to the incident light beam. Column eluate is nebulized with a relatively high flow

nitrogen or air and the mobile phase evaporated as the solvenugas mixture

passes down the vertically mounted evaporation chamber. At the bottom of the

chamber, all that is left

gas, solvent vapour and finely divided droplets or par-

ticles of analyte. This aerosol passes through the light beam and the photomul-

tiplier detects that portion of the incident light which

scattered by the analyte

(at an angle of 135’). At this angle, Charlesworth found the result to be effec-

tively independent of the

of the analyte.

The true linear working range of this instrument is not extensive, typically 1.5

orders of magnitude in concentration. Above and below this range, the size of

the analyte droplets produced no longer promote the reflection and refraction of

the light. Although this is a drawback, it is a relatively minor one, as the re-

References

pp.

72-3

Nebuliser

Chapter

I-

118

=Exhaust

Fig.

12.3.

Schematic

evaporative

mas

detector.

sponse functions due to Rayleigh and Mie scattering in the non-linear regions are

well described and still make calibration possible.

The chief disadvantage of this detector is the volatility limitations imposed

upon the analyte. The solvent evaporation chamber is, in effect a mild blow-

down apparatus which removes the mobile phase. If the analyte volatility

va-

pour pressure approaches that of the mobile phase it will vaporize and give no

response. In our laboratory, we have found hexadecane (b.p.

256°C)

to be par-

tially evaporated when the detector is operating at ambient temperature with

hexane (b.p.

68°C).

It is therefore likely that hydrocarbons below n-CI7 will not

give full recovery. Even given this limitation, the detector finds considerable use

for intermediate and low volatility analytes.

12.2.7.3

Dielectric constant detector

With few exceptions, the dielectric constant of a substance increases with its

polarity.

analyte elutes from the column, the dielectric constant of the

eluate will change. The dielectric constant of a non-polar or semi-polar sub-

stance is a function of its refractive index and as such many of the practical con-

siderations concerning

detectors apply equally to the

DCD.

more detailed

treatment may be found

Scott

[8].

HPLC

and column liquid chromatography

361

A typical DCD is a differential, temperature controlled concentric cylindrical

capacitor through which the column eluent flows. The cell electrodes are made

of stainless steel and are connected electrically as one side of a Wein or Schering

Bridge. If the mobile phase is less polar than the analyte, as in normal phase LC,

the dielectric constant of the eluate will increase as the peak elutes. The reverse

is usually the case in Rp-HPLC, and in order to avoid negative peaks in reversed

phase applications, DCDs allow for polarity reversal.

Setting up and balancing DCDs can be a tedious business as each side of the

bridge circuit needs to be balanced in an iterative fashion until the potential dif-

ference across the bridge is zero.

The linear dynamic range of the DCD is heavily influenced by the difference

in the dielectric constants of the mobile phase and the analyte, but has been

quoted as 3.5

lo4

which is comparable to the

detector.

12.3

QUANTITATION

In the majority of petroleum applications of HPLC, calibration is by external

standardization and quantitation is by peak area. Where samples are analyzed “as

received” or after dilution only, this approach is reliable and accurate. Where the

sample is worked up before analysis by liquidliquid or liquidsolid extraction, it

is necessary to determine the extraction efficiency (or “recovery”) in order to be

certain that a representative extract has been obtained. Where extraction effi-

ciencies are low or where time does not allow the recovery to be determined, an

internal standard or a standard addition method should be employed, provided

the detector response to the solutes is linear in the range

interest.

Peak area is most usually used for quantitation, as this is the most statistically

precise measure of analyte concentration. It does presuppose good resolution

however, and where this is not the case, a range of deconvolution methods or

even peak height measurement may have to be considered.

Contemporary HPLC now has a vast range of competitive quantitation devices

and statisticaVgraphica1 software available. Stand alone benchtop integrators,

microprocessor and PC data stations, local area networks (LANs), laboratory

information management systems (LIMS) and even mainframe chromatography

packages are all available. Selection is a compromise between cost, specification

and, increasingly, compatibility with existing computer hardware. Any

these

devices can take detector output and convert it to a high quality graphical or nu-

merical report, automatically labelled with peak identities according to previ-

ously recorded retention windows.

Caution is necessary, however, as any system will only act according to the

way it is configured by the operator. At each stage of the data systems applica-

References pp.

372-374