Adlard E.R. (ed.) Chromatography in the Petroleum Industry
Подождите немного. Документ загружается.
3
72
Chapter
12
A
mixed heavy end sample was separated into fractions with a
50
x
11
cm
silica gel Si60 column by Lancas
et
al.
[72].
Two grams of the sample were
mixed with silica in a precolumn and single solvents or binary/tertiary mixtures
used to fractionate it. Six solvents in six mixtures of increasing eluotropic
strength gave saturates, monoaromatics, diaromatics, triaromatics, polynuclear
aromatics, resins, asphaltenes and asphaltols. Typical recovery
is
quoted as bet-
ter than
90%
in most cases, with
an
RSD
of
1.2%.
12.7
FUTURE
TRENDS
A
number
of
more selective column mechanisms are beginning to find appli-
cation in petroleum analysis. Most specifically, the range
of
selectivities now
commercially available in normal phase charge transfer columns such as DNAP,
TNAO
and
TNAP
columns are allowing a more precise definition of aromatic
type cut point.
As
the industry has a continuing need for more precise total aro-
matic and aromatic type quantification, it is expected that the use of such col-
umns
will
increase. Similarly, the separation of functionalized “bad actors” from
a
range
of
hydrocarbon products may prove to be accomplished by anion and
cation exchange columns which are now also commercially available.
Undoubtedly, the single most useful advance
in
detector design for the petro-
leum industry has been that of the evaporative mass detector. This detector will
find increasing use
in
the field for
two
reasons. Firstly, the operation of the de-
tector necessarily results in volatile sample matrices being evaporated along with
the HPLC mobile phases used
in
petroleum applications. This may actually
prove to be an advantage in the analysis
of
some gasoline additives. Secondly, as
heavier products will not suffer the same fate as gasoline/naphthas, characteriza-
tion
of
such samples can take advantage
of
the detector’s true linearity and com-
position independence. These characteristics are unique in such a robust and
relatively inexpensive device.
Finally, LC-GC
is
still waiting for an enterprising manufacturer to develop a
truly turnkey system. Numerous applications of this hyphenated technique al-
ready exist which should be transferable. The analysis of oxygenates in gasoline
at percentage and at trace levels may yet prove to be the application which
arouses a sufficient volume of interest to be commercially viable.
12.8
REFERENCES
1
A.J.P.
Martin
and
R.L.M.
Synge,
J.
Biochem.,
35
(1941) 1358.
2
M.
Tswett, Proc.
Warsaw
SOC.
Natl.
Sci., Biol. Sect.,
14
(1903)
No.
6.
3
R.
Kuhn
and
E.
Lederer, Ber. Deut. Chem. Ges.,
64 (193 1)
306.
HPLC
and
column
liquid chromatography
3 73
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
L.R. Snyder and J.J. Kirkland, Introduction to Modem Liquid Chromatography, Wiley, New
York (1974).
N.A. Parris, Instrumental Liquid Chromatography, Elsevier, Amsterdam (1984).
M.T. Gilbert, High Performance Liquid Chromatography, Wright, Bristol(l987).
R.
Amos,
in Chromatography in Petroleum Analysis (Altgelt and Gouw, Eds.), Marcel
Dekker, New York. (1 979).
R.P. W. Scott, Liquid Chromatography Detectors, Elsevier, Amsterdam (1977).
A.T. James, J.R. Ravenhill and R.P.W. Scott, Chem. Inst., 746 (1974).
E.C. Homing, D.I. Carroll,
I.
Dzidic, K.D. Haegele, M.G. Homing and R.N. Stillwell, J.
Chromatogr. 99 (1974) 13.
R.P.W. Scott, C.G. Scott, M. Munroe and J. Hess Jr., J. Chromatogr., 99 (1974) 395.
P. Arpino, B.G. Dawkins and F.W. McLafferty, J. Chromatogr. Sci., 12 (1974) 574.
W.M.A. Niessen and J. van der Greef, Liquid ChromatographyMass Spectrometry, Marcel
Dekker, New York (1992).
S.D. Woodruff and E.S. Yeung, Anal. Chem., 54 (1982) 1174.
S.D. Woodruff and E.S. Yeung, Anal. Chem., 54 (1982) 2124.
D.L. Ford and W. Kennard, J. Oil Colour Chem. Assoc., 49 (1966) 299.
D.L. Ford and W. Kennard, Aust. Pat. Appl. No. 33406.
J.M. Charlesworth, Anal. Chem.,
50
(1978) 1414.
Standard Methods
of
Analysis and Testing
of
Petroleum and Related Products, 1993, Institute
of
Petroleum and Wiley, London.
E. Katz and K. Ogan, 5th Int. Symp. on Chem. Anal. Biol. Fate Polynucl. Aromat. Hydrocar-
bons, Battelle Press, Columbus,
OH
(1980) pp. 169-178.
J.
Grimalt and J. Albaiges, Afinidad, 40(385) (1983) 223.
B.A. Tomkins and W.H. Griest, J. Chromatogr., 386 (1987) 103.
P.R. Fielden and A.J. Packham, J. Chromatogr., 479 (1989) 117.
R.K. Symons and
I.
Crick, Anal. Chim. Acta, 151 (1983) 237.
T.
Saito, Bunseki Kagaku, 39 (1990) 21 1.
S.
Coulombe and H. Sawatzky, Fuel, 65 (1986) 552.
J.P.F. Palmentier, A.J. Britten, G.M. Charbonneau and F.W. Karasek, J. Chromatogr., 469
(1989) 241.
C.E. Ostman and A.L. Colmsjo, Fuel, 68 (1989) 1248.
J.R.
Mazzeo,
I.S.
Krull and P.T. Kissinger, J. Chromatogr., 550 (1991) 585.
M.C. Paputa-Peck, R.S. Marano, D. Schuetzle, T.L. Riley, C.V. Hampton, T.J. Prater, L.M.
Skewes, T.E. Jensen, P.H. Ruehle, L.C. Bosch and W.P. Duncan, Anal. Chem.,
55
(1983)
1946.
W.A. MacCrehan, W.E. May, S.D. Yang and B.A. Benner Jr., Anal. Chem., 60 (1988) 194.
K.
Jinno, J.C. Fetzer and W.R. Biggs, Chromatographia, 21 (1986) 274.
C. Borra, D. Wiesler and M. Novotny, Anal. Chem., 59 (1987) 339.
F.
Andreolini, C. Borra, D. Wiesler and M. Novotny, J. Chromatogr., 406 (1987) 375.
G. Grimmer, J. Jacob, K.W. Naujack and D. Schneider, Erdoel Kohle, Erdgas, Petrochem.,
38(2) (1985) 82 (Synopsis 8504).
L. Nondek and V. Chvalovsky, J. Chromatogr., 3 12 (1984) 303.
J.S. Thompson and J.W. Reynolds, Anal. Chem. 56 (1984) 2434.
S.J. Marshman, Fuel, 5 (1990) 364.
R.E. Rebbert, S.N. Cheder, F.R. Guenther and R.M. Parris, J. Chromatogr., 284 (1984) 211.
R.G. Christensen and E.V. White, J. Chromatogr., 323 (1985) 33.
W.A. MacCrehan and J.M. Brown-Thomas, Anal. Chem., 59 (1987) 477.
374
Chapter
12
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
J.C. Monin and R. Pelet, Adv. Org. Geochem. Proc. 10th Int. Meet., Wiley, London,
(1983)
G.E. Hayes and D.E. Hillman,
J.
Chromatogr.,
322 (1985) 376.
J.F. Schabron and M.P. Fuller, Anal. Chem.,
54 (1982) 2599.
K. Musha, C. Nagata and
S.
Tanaka, Bunseki Kagaku,
34(3) (1985)
T26.
K. Musha, C. Nagata and
S.
Tanaka, Bunseki Kagaku,
34(3) (1985)
T40.
F.P. DiSanzo, G.L. Johnson,
V.J.
Giarrocco and
P.
Sutton, J. Chromatogr. Sci.,
26 (1988) 77.
G.R. Bear, J. Chromatogr.,
459 (1988) 91.
E.N. Chen and V.P. Nero, J. Chromatogr.,
549(1-2) (1991) 247.
I>. Blanco-Gomis, J.M. LeBourgeois and R. Rosset, Chromatographia,
3 l(1-2) (1991) 71.
K.J. Welch and N.C. Hoffman, J. High Resolut. Chromatogr.,
15 (1992) 171.
J.
Apffel and
H.
McNair, J. Chromatogr.,
279 (1983) 139.
D.J. Cookson, C.J. Rix, I.M. Shaw and B.E. Smith, J. Chromatogr.,
312 (1984) 237.
F. Munari, A. Trisciani,
G.
Mapelli,
S.
Trestianu, K. Grob Jr. and J.M. Colin, J. High Resolut.
Chromatog. Chromatogr. Commun.
8 (1985) 601.
P.C. Hayes and S.D. Anderson, J. Chromatogr. Sci.,
26 (1988) 250.
J.F. Haw, T.E. Glass and H.C. Dom, Anal. Chem.,
55 (1983) 22.
I.L. Davies, K.D. Bartle, G.E. Andrews and P.T. Williams,
J.
Chromatogr. Sci.,
26 (1988)
125.
G. Felix,
E.
Thoumazeau, J.M. Colin and G. Vion, J. Liq. Chromatogr.,
10 (1987) 21 15.
I.L. Davies, K.D. Bartle, P.T. Williams and G.E. Andrews, Anal. Chem.,
60 (1988) 204.
R.N. Hazlett, H.C. Dorn, and T.E. Glass, NATO
AS1
Ser.,
Ser.
C (Magn. Reson.),
124 (1984)
709.
K.A. Caswell, T.E.
Glass,
M. Swann and H.C. Dom, Anal. Chem.,
61
(1
989) 206.
M.
Swann, Report,
BRDEC-2478,
Order No. AD-A
21 1879, 60
pp. Avail NTIS
(from
Gov.
Rep. Announce. Index
(US),
(1990) 90(1),
Abstr.
no.
000,691.
M.
Lienne, P. Gareil.
R.
Rosset, J.F. Husson, M. Emmelin and B. Neff, J. Chromatogr.,
395
(1987)
255.
C.S.
Hsu, M.A. McLean, K. Qian, T. Aczel,
S.C.
Blum, W.N. Olmstead, L.H. Kaplan, W.K.
Robbins and W.W. Schulz, Energy Fuels,
5
(1991) 395.
C.S.
Hsu, M.A. McLean, T. Aczel, S.C.
Blum,
W.N. Olmstead, L.H. Kaplan, W.K. Robbins
and
W.W. Schulz, Prepr.
Am.
Chem. SOC. Div. Pet. Chem.,
35 (1990) 649.
1993
Annual Book
of
ASTM Standards, Section
5,
ASTM
1916
Race Street, Philadelphia, PA
P.
Pei, J. Britton and
S.
Hsu,
J.
Liq. Chromatogr.,
6 (1983) 627.
P.
Pei and S.M.
Hsu,
J. Liq., Chromatogr.,
9 (1986) 331 1.
G.E. Fodor and F.M. Newman, ASLE Trans.,
22 (1979) 389.
Y.
Yamamoto and H.
Akutsu,
Bunseki Kagaku,
35 (1986) 23.
V.
DeSimone,
E.
Fogelqvist, R. Schneider and J. Pasanau, Toxicol. Environ. Chem.,
16
(1987)
51.
F.M. Lancas, E. Carrilho, G.H.N. Deane and M.C.F. Camilo, J. High Resolut. Chromatogr.,
12 (1989) 368.
pp.
839-846.
19103-1 187,
USA.
E.R.
Adlard
(Ed.),
Chromatography in the Petroleum Industry
Journal
of
Chromatography
Library
Series,
Vol.
56
0
1995
Elsevier Science
B.V.
All
rights
reserved
375
CHAPTER
13
Modern data handling methods
N.
Dyson
Dyson
Instruments
Lid., Hatton
Lyons
Industrial
Estate, Hatton,
Houghton-le-Spring,
Tyne
and Wear DH5
Om
UK
13.1 INTRODUCTION
The majority of current chromatographs use a data processor, either a stand
alone integrator or a computer-based system, to record, process and display
chromatograms and results. Prices range from just over
$1000
for a single chan-
nel instrument, up to many thousands
of
doIlars for multi-channel computer sys-
tems.
The electronic technology
of
signal processing has advanced with the micro-
processor. Report preparation and distribution using spreadsheets, databases and
electronic mail can be both stunning and delightful, but the theory of peak meas-
urement used within these data processors is still that of the
1960s.
Very little
of
the research on peak separation is incorporated into commercial data processors,
overlapping peaks are still separated by perpendiculars and tangents. This meth-
odology has severe limitations and really ought not to be applied to the meas-
urement of all peaks, but
it
is.
Current data processors are no more accurate than
a chromatographer of the
1960s
working with a pencil and ruler, but they are a
lot more precise and very convenient to use. They can deliver results
as
soon as
the analysis
is
completed and it is very easy for the unwary professional or nov-
ice analyst to accept these results at face value.
The purpose of this chapter is to indicate how measurement errors arise,
highlight the limitations of modern data handling methods in coping with these
errors, and point to strategies
of
method development that minimize the difficul-
ties
of
data processing. It will steer analysts towards the correct balance between
good and sufficient method development and realistic use of integrators and
computers. Chromatographic data processors are not a substitute for a column
or
References
pp.
398-399
376
Chapter
13
good experimental technique. Even the most expensive data processor is quite
capable of delivering a highly precise but totally inaccurate set of results.
13.2
THE
ROLE
OF
THE DATA PROCESSOR
13.2.1 The modern integrator
has
four distinct roles
(i)
Measure the detector signal.
To take the signal as it emerges from the
detector and prepare it, without distortion or other
loss
of information,
for processing. The detector signal
is
analog
in
nature and is sampled by
an analog to digital
(AD)
converter on the integrator input. This string
of sample readings
is
stored inside the integrator and processed by first
improving the signal to noise ratio and then extracting information
on
the size of a peak and its retention time.
Calculate and report solute information.
Analyte quantity and analyte
composition in terms of mass or concentration is measured from peak
area following calibration and the determination of response factors
from standard solutions. Height
is
an alternative measure of quantity
when the signaynoise ratio is poor. Analytes are identified by the reten-
tion time window in which they elute.
(iii)
Control the experiment.
Time-based programs control the mechanical
functions of the analysis: start the auto sampler, switch over valves, open
and close relays, etc. These programs can be in the chromatograph but
are often in the integrator. Similar programs optimize parameter settings
during the analysis. At the end of the experiment, the system is auto-
matically restored to its original configuration for the next analysis, or it
is re-configured for a different analysis.
(iv)
Monitor the analysis results for feedback control,
Additional informa-
tion from the integrator is used to monitor experimental stability. The
quality of the experiment can be assessed from additional data output:
(ii)
The coefficient of variation of results over
a
series of experiments is a meas-
ure of instrument control. Variations
in
retention time indicate a lack of control
of mobile phase flow rates, column temperatures
or
gas leaks. The total peak
area count is a measure of injection volume.
Its
constancy
(or
otherwise) is a
measure of injection proficiency. Column condition can be monitored by meas-
urement
of
the area and height of a large isolated peak, (see Fig.
13.1).
The ratio
(A/H)2
is
an approximate measure of peak variance
[
13
and therefore column ef-
ficiency. Peak asymmetry ratio,
BIA
[2],
defined in Fig.
13.2,
is a test of column
suitability. If too many peaks are asymmetric, it may be better to switch to a col-
umn
of more appropriate polarity.
Modern data handling methods
311
Fig.
13.1.
Whatever the peak shape, the ratio of its area
to
height is equal
to
its
width at some
height.
These are quantities that are readily calculated from existing data and can be
used as control diagnostics. The measurements can be used in
a
control feedback
loop to delay the next analysis until conditions are stable enough, or sound
an
alarm if they are too bad for too long. It is possible to build systems which allow
automatic carrier gas and column switching followed by purge cycles, these re-
Fig. 13.2. Asymmetry ratio
=
BIA
at 10% peak height [2].
References
pp.
398-399
378
Chapter
13
route samples to an alternative chromatograph if a fault occurs and nobody cor-
rects
the problem in a specified period.
13.2.2 Lab economics
The sum of all the analysis times gives the utilization of a chromatograph.
This is used to estimate spare capacity and decide whether an additional chroma-
tograph
is
required. Utilization divided by the instrument costs provides the
capital cost per analysis. The numbers of similar analyses performed by different
analysts may be compared to judge who are the most efficient workers.
13.2.3 Prepare and distribute data, information and reports
Even “stand alone integrators” are not always stand alone any longer. Micro-
processor developments and data communications software for chromatography
[3]
allows information generated by integrators to be incorporated into sophisti-
cated reports which show trends and make comments as well as give the basic
figures. These reports can then be distributed via electronic mail to a variety of
destinations.
In
some QC labs, the chromatographer’s role is to develop the
method and guarantee the quality of the analysis; he need not even see the
analysis results
or
be asked to comment upon them
[4].
13.3 LIMITATIONS
OF
THE
DETECTOR SIGNAL
It must not be assumed that the detector signal accurately reflects the profile
of the injected solute, yet it is clearly meaningless to measure the detector signal
unless it does. The signal is not representative of the solute unless the following
conditions apply.
13.3.1 Injector fractionation is uniform
Any injection port splitting of solutes must split each component species in
the same proportion as all others, i.e. there must be no discrimination. This sim-
ple requirement is very hard to achieve and has haunted the design of WCOT
injection ports. The multiplicity of existing designs implies that no single device
has
yet “got
it
right” for all types of solutes. Although improvements in injector
design have been made and are continuing, it still requires a competent analyst to
produce good quantitative results using WCOT columns,
Modern data handling methods
3
79
Splitting carrier gas streams into
two
columns or detectors usually creates
fractionation problems.
13.3.2 All solutes
of
interest have eluted
Columns sometimes retain part of the sample after the end of the analysis,
which is why many need reconditioning regularly. The peaks which purge during
reconditioning were possibly .weighed as part of an earlier analyte but, clearly,
they did not appear in the report of that analysis.
13.3.3 Column resolution is adequate
If solute peaks are not resolved sufficiently they cannot be measured accu-
rately. The transition from packed to capillary columns showed just how well
peaks can conceal other peaks. Increased column performance does resolve
peaks and presents them for measurement, but the analyst can reach a point
where there is too much resolving power, where a single component of interest
splits into its isomers and these have to be rounded up and added together again,
if they can be identified, to give the required result.
13.3.4 All solutes
of
interest are detected
Solutes that produce zero response from a detector cannot be measured: FIDs
cannot detect water or inorganic gases, for example. Selective detection
turns
lack of response into a virtue although the analyst must be very careful that his
quantitation does not deteriorate because some solutes are measured at the bal-
ance but not at the detector.
13.3.5 Signallnoise ratio is high enough
A
solute peak must be clearly distinguishable from the background noise.
ACS
guidelines
[5]
state that
a
peak cannot be clearly recognized unless the
S/N
ratio is at least
3:
1.
The mininium ratio for quantitation is 10:
1,
although at this
ratio, the results are not expected to be particularly good.
In practice, what chromatographers regard as a “noisy baseline” has a
S/N
ra-
tio
of
about
25
or
30:l
and there are theoretical reasons for defining a “good
baseline” as having a
S/N
ratio of
1OO:l
or better
[6]
because when the signal is
this clean, the data can be used. for peak modelling.
References
pp.
398-399
3
80
Chapter
13
Fig.
13.3.
Detector non-linearity effectively
folds
over the peak top.
13.3.6 Detector
is
linear
Saturation of the detector electronics is a built-in safety device to prevent the
electronics fiom damaging themselves. When the amplifier saturates, which it
can do on any of its range settings, it stops transducing the cell signal into an
output representative
of
the passing solute.
Before saturation, the relationship between analyte and detector signal
goes non-linear and gives a distorted measure of the solute. In the non-linear re-
gion, the peak top can be thought of
as
folding over (see Fig. 13.3). In extreme
cases
when the top folds over completely, it can appear that there are
two
peaks
Until all
of
these factors have been tested and their associated problems over-
come, all chromatograms must be considered suspect sources of information no
matter how repeatable the results are.
~71.
13.4
DETECTOR
SIGNAL
PROCESSING
13.4.1
Improving
signal to noise ratio
Detector signals always contain noise (unwanted frequencies) from chemical
and electronic sources. Noise hinders signal measurement and is therefore re-
moved before measurement begins. Detector sensitivity is increased by improv-
ing the signal to noise ratio
(S/N).
There are
two
basic techniques which data
processors currently employ: electronic filtering and software smoothing.
Modern data handling metho&
381
13.4.2 Electronic filtering
Electronic filtering uses inductive
(LR)
or capacitive (RC) components to ir-
reversibly filter high frequency noise. Inductive components are used in mains
power filters, capacitive components remove instrument generated noise. Time
constants are matched to the noise they are intended to remove. If they are too
small, they will leave
too
much noise, if they are too large they will remove the
noise but they will also distort peak shape; retention time and peak height are not
then reliable measures but the peak area remains constant. This distortion can
manifest itself in capillary chromatography where an old GC is used with a high
resolution column. The time constant of the detector amplifier may have been
built to accommodate packed column peaks and be too large for the narrower
WCOT peaks which are distorted in consequence.
13.4.3 Software smoothing
Software smoothing uses mathematical operations to remove noise. These al-
gorithms are either based on the moving window functions described by Savitsky
and Golay
[8,9]
or they use software filters matched to the noise fiequency
[lo].
Smoothing is performed on the: stored signal data and, unlike electronic filtering,
is reversible as the original data is not destroyed. Moving window smoothing
assumes
a
high enough data sampling rate (about
100
samples per peak
[
1
l]),
and not too much peak asymmetry otherwise the peak shape is further distorted
[12].
Even when sampling is fast enough, Savitsky4olay smoothing has the
disadvantage, that unless the noise is random, it moves high frequency noise into
the low frequency domain where it looks like baseline rubble and can interfere
with peak location or be measured in error
as
peaks
[
131.
At high
S/N
ratios, all smoothing techniques work well but trace analyses are
important, and this is where smoothing
is
least effective and most needed.
Smoothing algorithms are built into the integrator by the manufacturer and vary
from one manufacturer to the next. They are unknown to the analyst and their
validity cannot be easily checked.
13.4.4 Measurement of peak areas
One of the results of the rapid development of microprocessor-based instru-
ments is that all analysis data are stored and all signal processing is now post-
run, with the benefit of “hindsight” calculation. Data processors store the detec-
tor signal in digitized form.
A
single chromatographic peak
is
measured by the
References
pp.
398-399