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

Подождите немного. Документ загружается.

362

Chapter

tion, the user must be certain that each setting is sound in order to obtain a final

quantitative output of the highest possible integrity.

12.4

APPLICATIONS

The applications of HPLC in petroleum analysis are summarized in Table

12.1. The wide variety of separation mechanisms, column chemistries and de-

tection systems represented by HPLC offers the petroleum chemist a range of

distinct systems. In general these fall into three categories:

(1)

the separation and direct quantitation of individual compounds;

(2) separation and characterization of compound classes such as, for example,

(3)

preparative or semi-preparative fractionation of complex mixtures for de-

saturates, olefins and aromatics in petroleum products;

termination by other analytical techniques.

Within each category, standard methods exist for particular determinations,

which have been rigorously tested in terms of inter-laboratory precision. Such

standard methods as exist within the Institute of Petroleum handbook, Standard

Methods

Analysis and Testing

Petroleum and Related Products,

1993

[19]

are discussed

the following sections.

12.4.1

Individual

compounds

12.4.1.1

Polycyclic aromatic hydrocarbons

(PAHs)

These compounds have attracted considerable interest due to their role as

pollutants and, in some cases, their carcinogenic properties. Amos

[7]

cites some

early

WLC

applications. Katz and Ogan [20] have used partition and size ex-

clusion columns in series to effect the analysis, and a combination of normal

phase amino and reversed phase C 18 columns has been used to determine PAHs

in crude oil by Grimalt and Albaigks [21]. Further LC-LC methods, chiefly

aimed at benz[a]pyrene, have been employed by Tomkins and Griest

[22]

and

Fielden and Packham [23]. In the former case, Partisil silica and analytical scale

Vydac 201TP reversed phase columns were used and in the latter case cyclodex-

trin and

ODS

silica. In both cases, the selectivity and sensitivity of fluorescence

detection was used to determine the PAH directly.

Symons and Crick [24] have determined PAHs in refinery effluent after clean-

and preconcentration using a Radial-Pak

CIS

column with

75:25

aceto-

nitrilelwater eluent and

and fluorescence detection. Recoveries were vari-

HPLC

and column liquid chromatography

363

TABLE

12.1

APPLICATIONS

HPLC

PETROLEUM ANALYSIS

Crude oil

ha-arenes

Dibenzothiophene

Phenols

Polynuclear aromatic hydrocarbons (PAHs)

Preparation

PAH fractions

Saturates/aromatic types

Naphthdgasoline

Aromatic nitrogen compounds

Benz[a]pyrene

Saturateshromatic types

Saturates fractions

Saturates/olefindaromatics

Aviation fuel

Aromatic nitrogen compounds

Coumarin

PAHS

Saturateshrornatics

Saturateshromatic types

Saturates/olefindaromatics

2,4-Dimethyl-6-tertiarybutylphenol

DieseUdistillate fuels

Alkyl nitrates

Aromatic nitrogen compounds

Mono/di/triaromatics

Olefins

PAHS

Phenalenones

Saturatedaromatics

Saturatedaromatic types

Grimmer

et al.

Rebbert

et al.

Christensen and White

MacCrehan and Brown-Thomas

Grimalt and Albaiges

Ostman and Colmsjo

Welch and Hoffman

Nondek and Chvalovsky

Tomkins and Griest

Apffel and McNair

Cookson

et al.

Munari

et al.

Hayes and Anderson

ASTM

2002,2003

ASTM D

1319

Nondek and Chvalovsky

374

Fielden and Packham

IP PM-AT

Welch and Hoffman

Cookson

et al.

Hayes and Anderson

Haw

et al.

Davies

et al.

ASTM

D1319

Hayes and Hillman

343

Schabron and Fuller

Nondek and Chvalovsky

Lienne

et al.

Fielden and Packham

Davies

et al.

Marshman

Apffel and McNair

Cookson

et al.

391,

PM-AY

IP PM-AZ

References pp.

72-3

TABLE

12.1

(continued)

Chapter

Diesel exhaust particulates

Naphtho[8,1,2-abc]coronene

Nitrated PAHs

Fuel

oil

Benz[a]pyrene

Lubricating

oils

Additives (over

50)

Aromatichon-aromatic fractions

Renz[a]pyrene

Furfural

Naphthalene/phenanthrene

PAH

fractions

Polychlorinated biphenyls

Saturates/aromatics

Saturatedaromaticsipolars

Sulphonates

Sulphurized alkylphenols

improver

Zinc dialkyldithiophosphates

Heavy

oils

Olefinic fractions

PAH

fractions

Saturates/aromatics/PAH/resins/asphaltenes,

etc.

Saturates/naphthenes/alkylaromatics/thiophenes

Bitumen

PAH fractions

Refinery effluent

PAHs

Davies

al.

Hazlett

Jinno

al.

Paputa-Peck

al.

MacCrehan

al.

Tomkins and Griest

Musha

al.

Musha

al.

ASTM D

2549

Saito

DeSanzo

al.

Mazzeo

al.

Palmentier

al.

Ostman and Colmsjo

DeSimone

al.

368

ASTM

D2007

Pei

al.

Pei and Hsu

Bear

ASTM

D3712

Chen and Nero

Blanco-Gomis

al.

Fodor and Newman

Yamamoto

and

Akutsu

Coulombe and Sawatzky

Lancas

al.

Hsu

al.

Hsu

al.

Coulombe and Sawatzky

Symons and Crick

HPLC

and column liquid chromatography

365

able and less than 87% for

4-6

ring PAHs. Saito [25] determined benz[a]pyrene

in lubricating oils and greases with fluorescence detection after an alumina

clean-up; precision was reported as 67% RSD with a detection limit of

3.95 nglg.

HPLC has also been used purely as a fractionation technique for PAHs. Cou-

lombe and Sawatzky [26] applied this method to bitumens and heavy oils and

determined PAHs in the various LC fractions by GC. Palmentier

al.

[27] em-

ployed

semi-preparative scale fractionation followed by GC-MS. Ostman and

Colmsjo [28] prepared PAH fractions from crude oil and used crankcase oil by

elution from short silica columns followed by an automated backflushed Bonda-

pak-NH, HPLC system. Individual PAHs in the final fractions were quantified

by GC. Detection limits were in the order of

ppm from a 10-15 mg sample

using GC-FID or 0.1 ppm by scaling up the initial silica clean-up.

Mazzeo

et-

al.

[29] detected PAHs as quinones by oxidizing them with CeN.

Reductive mode electrochemical detection was employed to achieve detection

limits in the order of ppb. Chromatography was performed on an ODS column

using propan-2-01 /phosphate buffer as eluent. These authors applied the above

system to the analysis of naphthalene and phenanthrene in a motor

oil.

proposed Institute of Petroleum standard method,

PM-BN, also exists for

the determination of PAHs in petroleum, coal and shale oil products. Detection

limits of 0.1% dm of total, and 0.1 mg kg-1 of individual PAHs are quoted. The

method uses open, gravity feed silica columns to produce

PAH extract which is

further separated by HPLC on a 5,um particle Spherisorb amino column or

equivalent. The isolated

4-6

ring fraction is then run on a Sephadex LH20 parti-

tion column in order to separate alkylated PAHs from the parent PAHs. These

parent PAHs are individually determined by GC. Precision has yet to be estab-

lished.

12.4.1.2 Other indigenous compounds

Nitrated PAHs in diesel engine exhaust particulates have been examined by

Paputa-Peck

al.

[30] and MacCrehan

al.

[31]. Paputa-Peck employed a

normal phase HPLC fractionation of methylene chloride extracts. Determination

of individual nitrated PAHs was by GC with a nitrogen-phosphorus detector or

by GC-MS. MacCrehan separated methylene chloride extracts by RP-HPLC and

compared voltammetry, amperometry and fluorescence for direct detection of

individual compounds. Diesel particulates have also been examined by Jinno

al.

[32] for

naphth0[8,1,2-abc]coronene

using reversed phase separations and

multichannel

detection.

High molecular weight heterocyclic nitrogen and sulphur compounds have

been studied by Borra

al.

[33] and Andreolini

al.

[34]. These authors used

References pp.

72-3

366

Chapter

highly efficient capillary LC columns and a combination of direct diode array

fluorescence detection

fraction collection and mass spectrometry to examine

solvent refined coal distillate, syncrude and shale oil samples.

Polycyclic aromatic nitrogen compounds (aza-arenes) in Arabian Light crude

oil have been examined and identified by Grimmer

al.

[35].

These authors

used

TLC/HPLC isolation scheme with separation and identification of indi-

vidual compounds by GC. Aza-arenes, anilines and alkyl aromatic amines in

gasoline, kerosene and diesel fuel have also been studied by Nondek and

Chvalovsky

[36]

using

two

different charge transfer columns,

3-(2,4-

dinitrobenzene su1phonamido)propyl silica and

3-(2,4-dinitroanilino)propyl

sil-

ica. A comparison of five different charge transfer columns for the separation of

aromatic compound classes from a crude oil distillate sample and other fossil

fuel samples has been made by Thompson and Reynolds

[37].

Phenalenones such

7H-benz[d,e]anthranen-7-one,

benzanthrone and methyl

phenalones have been quantified in middle distillates by Marshman

[38]

using

silica reversed phase separation and

detection at

400

nm. Detection limits

quoted are typically

the region of

0.2

I-*.

Dibenzothiophene has been

quantified in crude oils by Rebbert

al.

[39]

and by Christensen and White

[40].

The former authors employed HPLC to fractionate samples for GC/FPD

quantification. In contrast, the latter authors used

novel LC-tandem MS system

to separate and unambiguously identify dibenzothiophene directly. Indigenous

phenols in crude oil have been examined by MacCrehan and Brown-Thomas

[41]

with detection limits of less than

100

ng/g. These authors used alkaline

sol-

vent extraction of the oil, solid phase purification of the extract and RP-HPLC

with electrochemical detection.

HPLC has even been applied to asphaltenes

order to assist the determina-

tion of an “average molecule. Monin and Pelet

[42]

used size exclusion and a

range of bonded phase columns to fractionate such samples after selective disso-

lution

a number

solvents.

12.4.

Additives

and contaminants

The anti-oxidant

2,4-dimethyl-6-tert-butyl

phenol has been quantified by nor-

mal

phase isocratic HPLC with

detection and is the subject of an

Standard

Method,

IP343.

The method allows a number of columdmobile phase combina-

tions and, in our experience, is robust and precise. Published repeatability and

reproducibility at the

200mg/l

level are

2.61

and

6.56,

respectively. Some

homologues and isomers of this compound may also be separated using varia-

tions in mobile phase composition. The same compound has also been quantified

with electrochemical detection by Hayes and Hillman

[43].

Alkyl nitrate cetane improvers in diesel fuel have been determined by

Schabron and Fuller

[44].

Normal phase LC on silica coupled with variable

HPLC

and column liquid chromatography

367

wavelength IR detection was used to separate and quantify amyl, hexyl and octyl

nitrates. Recovery, accuracy and precision quoted were good and detection limits

of 0.05 and 0.01 vol.% are given for amylhexyl and octyl nitrates.

Up to 50 lubricating oil additives have been separated and retention times de-

termined by Musha

al.

[45,46]. These authors used both normal and reversed

phase columns with

detection at the maximum absorbance for each com-

pound.

Furfural has been determined in lubricating oils by Di Sanzo

al.

[47].

5 pm ODs-silica column with an eluent of

70:30

watedmethanol and

detec-

tion at

280

nm gave a recovery >95%, good precision, and good agreement with

a bisulphite extractionKJV method. Samples for HPLC were pre-extracted with

methanol and cleaned-up with a C18 silica cartridge prior to determination.

Synthetic and indigenous sulphonates, including alkyl benzene sulphonates, have

been separated and quantified by Bear [48]. This author evaluated the evapora-

tive light scattering detector in the analysis of a wide range of surfactants and

concluded

“

uniform linear response for each class of surfactant, with detec-

tion limits in the low nmole range”. In particular, the detector response was re-

ported to be “independent of the alkyl chain length and the degree of aromatic-

ity” with respect to alkyl benzene and alkylaryl petroleum sulphonates. Columns

and mobile phase varied according to the application and samples were analyzed

after dilution of the parent product.

diode array

detector was also used in

series with the

ELS

detector. Standard deviations of all the analytes were less

than 1%.

Sulphurized alkylphenols have been separated from reaction side products and

base oil on a normal phase, y-cyclodextrin silyl column with gradient elution and

evaporative light scattering detection by Chen and Nero [49]. Individual fraction

from the separation were also characterized by mass spectrometry. Fingerprint

comparison between samples which passed and failed engine test specifications

are presented. The advantages of the ELSD over

detection were stated by

these authors to be freedom from ambient temperature variation effects, minimal

baseline drift with multiple solvent gradients and a response which was mass

dependent rather than concentration dependent.

To illustrate the breadth of HPLC applications in the field of lubricating oil

additives, normal phase and reversed phase methods have even been applied to

the characterization of

poly(styrene-alkylmethacry1ate)co-polymer

viscosity in-

dex improvers of molecular weight up to

300

000 Da by Blanco-Gomis

al.

Institute of Petroleum method exists for the determination of the coumarin

content of kerosene. This compound, 1,2-benzopyrone is often added to kerosene

as a marker for excise purposes. The method uses a silica column, a mobile

phase of

propan-2-01 in hexane or heptane and

detection at 274nm.

1501.

References pp.

3 72-3 74

368

Chapter

Typical calibration range is 2-4 mg/l and at the

mg/l level, repeatability

quoted as

0.06

and reproducibility

0.28.

12.4.1.4

Compound

classes

The inherent normal phase separation mechanism (adsorption) has the ability

to separate complex mixtures of hydrocarbons according to degree of unsatura-

tion.

such, it has been widely exploited in the characterization of petroleum

products with respect to the saturate, monoaromatic, diaromatic, triaraomatic,

polar and (to a lesser extent) olefin content. Ongoing development of bonded

normal phases has largely been aimed at achieving cleaner “cut-offs” between

compound classes, most notably by the use of substituent groups which separate

by charge transfer mechanisms with the aromatic nuclei of the sample compo-

nents. Products are often quantified

terms of the mass or volume fraction of

each compound class present, and further separation of individual components

within any class

either not possible or unnecessary.

fewer than five standard

methods of this type exist covering aviation

fuel,

auto diesel, drilling mud oils, gasoils and lubricating oil base stocks. Two

distinct HPLC technologies and quantitation methods are employed, both with

isocratic elution.

Silica columns and backflushing are used to separate saturates from total aro-

matics in basestocks (IP

368)

and gasoils

(IP

PM-AZ) with gravimetric quantita-

tion and in aviation fuel

(IF’

PM-AT) with

RVUV

detection and quantitation

total aromatics and naphthenes. In all three cases, saturates elute through the

columns unretained and aromatics (with or without olefins) are backflushed off.

auto diesel and drilling mud oils,

two

amino bonded phase columns are

used to separate mono, di and triaromatics with

detection and external stan-

dard quantitation (IP

391

and

PM-AY). The main concern in these last two

methods

that the external standards chosen are individual compounds, whereas

the actual sample components present

each class are many and varied. Detec-

tor response factors between sample and standard can therefore vary and will be

composition dependent.

wide range of petroleum products and crude oil have been characterized by

HPLC and it

best to consider each one in order of product type.

Crude oil has been characterized by Welch and Hoffman

[SO].

These authors

used an on-line microbore LC-GC-MS system with a 2,4-dinitrophenyl mercap-

topropyl silica LC column. The system employed a retention gap between the LC

and the GC columns and no attempt at quantitation was made. This article also

includes the analysis

JP-4

aviation fuel, isolating and identifying alkylben-

zenes, alkyltetralins, alkylbiphenyls, naphthalene and dimethyl naphthalenes.

Gasolines have been characterized by Apffel and McNair [52], Cookson

al.

[53],

Mussi

al.

[S4]

and by Hayes and Anderson

[SS]

on an aminopropyl silica

HPLC

and column liquid chromatography

column to separate alkene-free gasolines into saturates, monoaromatics, diaro-

matics, tri/polyaromatics and polar groups. Each group was quantified by

detector into weight percentage abundance. Calibration samples were obtained

by fractionation of a fuel sample rather than by use of single pure compounds in

an effort to minimize compositional and

response factor differences.

Both Apffel and McNair and Munari

al.

used on-line HPLC-GCRID meth-

ods to analyse gasoline saturates, unsaturates, aromatics and polar compounds.

The latter authors employed a retention gap between the two chromatographic

systems and microbore HPLC columns. Hayes and Anderson used off-line

HPLC

with a dielectric constant detector to achieve an accurate group type sepa-

ration and quantitation of gasoline with uniform response factors from the detec-

tor. The mobile phase was 2,2-dichloro-1,

1-trifluoroethane (Envron 123). The

individual fractions were then analyzed by GCMSD to identify components and

GC/FID to quantify them. The authors reported that spent Envron 123 can be re-

used several times without purification or easily redistilled on

continuous

basis.

Kerosenes have been characterized also by some of the authors previously

cited [51,53,55].

addition, Haw

al.

[56]

used a propylamino silica column

with on-line

NMR

the detector.

this case, the mobile phase was

l,l,l-

trichlorotrifluoroethane with

2.5%

deuterochloroform and

0.05%

hexamethyld-

isiloxane as

NMR

reference. Each compound class (monocyclic and dicyclic

aromatics) could be given an average composition. The average composition of

the saturate fraction was, however, limited by problems in accounting for qua-

ternary carbon. Davies

al.

[57] utilized the LC-retention gap-GCRID ap-

proach to a kerosene sample with microbore amino and silica glass-lined LC

columns in series with pentane eluent and backflushing. Unfortunately, the low

dead volume of detectors required for microbore LC precluded conventional

dielectric constant detectors and thus direct quantitation of the saturate and

aromatic fractions prior to GC was not possible. The system was automatic and

clearly improved the analysis of the aromatic fractions.

Diesel and distillate fuels have been studied by all the methods described for

crudes, gasolines and kerosenes [52,53,57]. Silica and amino columns have been

used to separate diesel into saturates, olefins and aromatics with

and/or

quantitation by Felix

al.

[58].

Davies

al.

[59]

used the LC-GC technique

previously described, but with specific reference to polynuclear aromatics in die-

sel fuel. The chromatographic system described by these authors produced a

complete fractionation by compound class but in this study emphasis was placed

on the definition of a two-dimensional (LC versus GC) retention map for se-

lected PAHs. By comparison, the retention indices of aromatic compounds from

the diesel sample led the authors to conclude that naphthalene, phenanthrene and

their alkyl derivatives were the predominant aromatics present.

Refeences pp.

3 72-3 74

Chapter

The

LC-NMR

approach, previously applied to kerosene, has been extended to

diesels and distillates by Hazlett

et al.

[60],

again culminating in the determina-

tion of average composition for each compound class. From LC-NMR data,

Caswell

et al.

and Swann found it possible to predict the physical properties of

diesel and jet fuels

[61,62].

Multiple regression analysis was used for the corre-

lation of

LC-NMR parameters from each fuel with

physical properties

such as cetane number, distillation data, flash point, pour point, density, etc.

Thirteen of the

properties in reference

[61]

had correlation coefficients in

excess of

90%

and seven were in excess of

95%.

Copper(li) and silver-modified silica columns have been prepared by passing

ammoniacal CuSO, through the column or by use of ammoniacal AgN03 during

packing by Lienne

al.

[63].

With pentane or Fluorinert FC72 as mobile phase,

olefins could be separated from light and heavy distillates with

and

de-

tection.

Heavy hydrocarbons have been characterized by Hsu

et al.

[64,65]

by on-line

LC-MS. It was reported that distinction could be made between naphthenoaro-

matics and alkylaromatics and also between aromatic hydrocarbons and thio-

phenes. The value of this kind of information for refinery processing is very

high.

12.5

PREPARATIVE

HPLC

AND

COLUMN

LIQUID

CHROMATOGRAPHY

12.5.1 Standard methods

The

1993

Annual Book of

ASTM

Standards

[66]

published by the American

Society for Testing and Materials lists six liquid column chromatography meth-

ods

of relevance to the petroleum industry.

ASTM

D13

is identical to the Institute of Petroleum, London method IPI

entitled

Hydrocarbon Types in Liquid Petroleum Products

Fluorescent Indi-

cator Absorption.

It is limited to samples boiling below

3 15°C

which are sepa-

rated by it into saturates, olefins and aromatics by elution through a silica col-

umn with 2-propanol under air or nitrogen pressure. Fluorescent dyes are added

to the top of the column which co-elute with the olefins and aromatics and serve

to mark the boundaries of each zone. The saturates front coincides with the wet-

ted front

the material passing down the column. The lengths of each zone are

measured at the end

the separation and these lengths are proportional to the

percentage of each class present in the sample.

This test has been in use with

slight modifications for many years and

especially relevant to gasolines and

aviation kerosenes. Its main drawbacks are that it is time-consuming and opera-

tor intensive and that strict control

the silica gel quality

critical.

HPLC

and column liquid chromatography

371

ASTM D2002 and D2003 are also silica gel fractionation methods used to

provide representative saturate fractions from low and high olefinic naphthas,

respectively. Again 2-propanol is used under pressure as the mobile phase.

ASTM D2001 uses

dual column system of Attapulgus clay and clay/silica

gel. Saturates, aromatics and polar fractions from oils with initial boiling points

greater than 260°C are recovered from the columns. This analysis

often re-

ferred to as the “Clay-Gel” method. Once again it is a time-consuming and la-

bour intensive technique. This type of separation has been the subject of

HPLC

development by Pei

al.

and by Pei and Hsu [67,68].

ASTM D2549 furnishes aromatics and non-aromatics for further analysis by

mass spectrometry. In this method, up to 10 g of oil boiling between 232” and

538°C

are separated on a column

bauxite and silica gel. Pentane

used to

elute the non-aromatics and the aromatics are eluted with successive portions of

diethyl ether, chloroform and ethanol.

ASTM D3712 uses a silica gel column with chloroform and ethanol to sepa-

rate diluent oil from petroleum and synthetic sulphonates. The average molecular

weight of the sulphonate is then determined by ashing the recovered material.

Metal sulphonates are first hydrolysed to sulphonic acids and converted

so-

dium sulphonates prior to the separation. Incompletely hydrolysed samples

not separate well and, as the analysis is again labour intensive, care must be

taken in the hydrolysis step to avoid

considerable waste of effort. This method

is identical with

Method 369.

12.6

INDIVIDUAL PUBLICATIONS

Zinc dialkyl dithiophosphate additives (ZDDPs) in finished lubricating oils

have been determined on 183 cm

0.78 cm glass columns packed with 37-

75pm silica by Fodor and Newman [69]. The alkyl/aryl ratio of the ZDDPs in

the recovered methylene chloride fraction was determined by

spectroscopy

against calibration solutions of the pure compounds. The authors concluded that

losses incurred in the chromatography limited the method to that of an estima-

tion rather than a quantitative determination.

Yamamoto and Akutsu used a preparative scale silica gel column to separate

saturates and aromatics and a 60

silica/lO-20% w/v silver nitrate column to

separate saturates from alkenes. The argentation column could produce a

fraction of alkenes from a column packed with 80 g of argentated silica. Typical

samples included heavy distillates from thermal cracking [70].

De Simone

al.

[71] obtained polychlorinated biphenyl extracts from petro-

leum samples by means of a combination of gel permeation and silica microcol-

umns. The PCB concentration in the extract was determined by

with electron

capture followed by mass spectrometry

confirm structures.

References

pp.

372-374