Adlard E.R. (ed.) Chromatography in the Petroleum Industry
Подождите немного. Документ загружается.
72
Chapter
3
The accurate control
of
the carrier gas by
a
back pressure regulator
or
mass
flow controller is also important
for
reproducible peak retention times. Flow
controllers have been strongly advocated for HTGLC analysis to theoretically
give an even resolution over the whole HTGLC temperature program range.
No
account was taken of the substantial increase in gas viscosity (37% for helium
between 150 and
250"C),
or changes
in
optimum velocity over the wide tempera-
ture range.
Also
mass
flow
controllers output constant volumes at ambient tem-
perature which expand at higher temperatures, considerably increasing the gas
velocity. The author has always thought that pressure control
is
correct for wax
chromatography. At constant pressure, the gas velocity should slowly decrease
with increase
in
temperature
so
as to keep the resolution of the n-alkanes rea-
sonably constant (providing the optimum velocity and temperature program rate
was chosen)
[32].
In
1990, Grob and Tschor [63] investigated the average carrier gas velocities
producing maximum separation efficiency at 370°C. They used a 25 m
X
0.32
mm i.d. column coated with
0.1
5
pm
of
a methyl-polysiloxane to compare
the van Deemter curves at 370°C and 60°C. Their results show that the optimum
gas velocities decrease with increase in temperature (approximately 35% be-
tween
60°C
and 370°C).
Also
the difference in optimum gas velocities of hydro-
gen and helium is negligible for the columns normally used for wax separation
(around
20
cm/s at 370°C). They proved that, at constant pressure, the decrease
in
gas velocities with increasing column temperature coincides with the decrease
in
optimum velocities, whereas, at constant flow, the gas velocity increases away
from the optimum. Therefore constant pressure controllers are preferable for
wax analysis, and helium or hydrogen can be used.
The choice
of
column dimensions for quantitative analysis depends on the
wax being analysed and the degree of resolution required for accurate quantita-
tive analysis. Decreasing the column diameter from 0.53 to
0.10
mm greatly in-
creases the number of effective plates per metre (i.e. narrow columns have far
greater separating power). However, the capacity of the column decreases with
decrease
in
internal diameter,
so
that sample size must be reduced.
To
obtain
high resolution at high temperature a thin stationary phase film
is
used
(0.1-
0.25
,pni
thickness); this further reduces the column capacity. The length
of
the
column affects the analysis time and the resolution. If the column is too long, the
higher molecular weight components suffer from peak broadening and eventu-
ally are undetected.
Short
wide bore columns
(8
m
x
0.53
mm i.d.), made either of stainless steel
or
aluminium-clad silica, can be used for Polywax separation. However, this size
of
column
is
inadequate for resolving other synthetic waxes such as the Fischer-
Tropsch waxes. These waxes require the resolution
of
a
15
m
X
0.32 mm i.d.
column. The new stainless steel columns are probably very appropriate for sepa-
The chromatographic analysis
of
refined and synthetic
waxes
73
rating these materials. Paraffin waxes and most wax blends can be efficiently
separated by a high quality 25 m
X
0.32 mm i.d. polyimide-clad column. Wax
blends with large amounts of high melting point wax require a 0.25-0.20 mm
i.d.
column, 15-25 m long
to
obtain adequate resolution at high temperature. High
melting point microcrystalline waxes require high temperature microbore col-
umns to resolve
all
the n-alkanes. The separation
of
a microcrystalline and syn-
thetic wax
is
shown in Fig.
3.4.
30
110
20
b
C30
I
40
50
60
Mins
1
C40
30
40
50
60
Minutes
Fig.
3.4.
(a) Capillary GLC separation
of
synthetic
wax.
(b)
Capillary GLC separation
of
microcrys-
talline
wax.
Chromopak
15
m
X
0.15
mm
HT,
temp. prog. 6W20°C, 6°C mid.
References
pp.
9&93
74
Chapter
3
For accurate quantitative analysis of both synthetic and refined waxes, a high
degree of resolution
is
required. This
is
not only obtained by the choice of col-
umn parameters and carrier gas flow, but also from the temperature program
conditions selected. With modern
GC
instruments, it is possible to accurately
control all the column oven conditions to obtain optimum resolution. Wax chro-
matography requires a slow temperature program to obtain the required resolu-
tion; this sometimes leads to analysis times of over
1
h.
The initial temperature of the GC oven is governed by the solvent used, and
should be set low enough to condense the sample solution in the column en-
trance.
An
initial short isothermal period is required for injection and the forma-
tion
of
a narrow sample band in the column. The oven can then be rapidly pro-
grammed (20-3O0C/min) to a temperature just below the first component’s elu-
tion temperature. This varies according to the sample, the column used and the
carrier gas flow rate. Another short isothermal period stabilizes the column tem-
perature before the sample separation stage of the temperature program.
For quantitative analysis, the separation stage has to be
slow
enough to
resolve the n-alkane peaks as near as practicable to the baseline.
A
tempera-
ture program rate between
4
and 8°C is commonly used. Ideally the final
oven
temperature is set
so
that the last peak has, if possible, eluted [13], but
if
this is not feasible, then the final isothermal period should not be much
longer than approximately 5 min, otherwise the final peaks will broaden and
merge with the baseline. The elution temperature of individual hydrocarbons
depends
on
the ratio of the column radius to the film thickness (i.e. the phase
ratio),
and the column length. This is the reason behind the choice of a short,
wide bore, thin film column for the high temperature separation of high mol-
ecular weight synthetic waxes. (Although narrow bore, thin film high tempera-
ture columns are required to resolve microcrystalline waxes for quantitative
analysis.)
Between 1985 and 1987, the author carried out research into the carbon
number distribution
of
wax blends
[13,32].
This showed the importance
of
knowing the response factor of alkanes of higher carbon number, as they vary
depending on injection technique and chromatographic separation conditions.
Several authors have used pure alkanes up to C44 to illustrate the differences in
response factor of different injection techniques. However, only a few authors
[
25.62,64] carrying out simulated distillation research have clearly shown that
pure hydrocarbons can be used to measure alkane response factors up to C87.
Pure alkanes up to C60 [65] can now be purchased, which means that
GC
equipment can be calibrated for response factors for the analysis of most refined
waxes. Polywaxes are used to empirically calibrate equipment for higher carbon
number responses, but there
is
a need for an internationally recognizable stan-
dard for HTGLC.
The chromatographic analysis
of
refined and synthetic waxes
75
3.2.3.
I
Carbon number distribution analysis
Having successfully separated the wax, the next problem is to accurately
analyse the complex chromatogram. This type of quantitative measurement is
commonly
known
as carbon number distribution analysis. It sets out to quantify
the amount of each straight-chain (or normal) alkane, as well as the groups of
branched (or iso-alkanes) and cyclic alkanes
of
the same carbon number. For
each carbon number, this is simplified by measuring the peak area of each pre-
dominant normal alkane and separately totalling the preceding smaller peaks
back to the previous n-alkane to give the theoretical iso-alkanes of the same car-
bon number
[40].
The results can then be plotted to give straight, branched and
total carbon number distribution graphs. The ratio of straight to branched chain
alkanes can also be obtained which is characteristic of the particular type of wax.
Wax chromatograms are complex and many
GC
methods do not resolve peaks
down to the baseline. Therefore, there are several ways in which the peak areas
can be measured. The theory and general practice of integration is covered else-
where
[66],
but a more detailed explanation is required to carry out accurate car-
bon number distribution analysis. Historically, the development of this type of
analysis has arisen from a need by rubber manufacturers for a detailed specifica-
tion for wax blend additives (using packed columns). The lack of both chroma-
tographic resolution and a reproducible
GC
method made interlaboratory agree-
ment difficult. This could only be accurately achieved by using a reference col-
umn and wax blend,
so
that the laboratories concerned could adjust the
GC
conditions to obtain the same resolution. Also the method of integration had to
be clearly stipulated to obtain reproducible results. The three main methods of
packed column integration, which are the basis for capillary
GC,
are skim base-
line (following only the main troughs), valley-valley baseline (following every
trough) and horizontal baseline
[
131.
Barker investigated the accuracy of these different methods of integration us-
ing a standard rubber wax blend containing paraffin and microcrystalline waxes
[
13,321.
Using a modified form of the I.P. Method
372/85,
the valley-valley and
skim baseline integration results were found to be similar, but the horizontal
baseline results differed markedly. It was noted that the baseline hump was con-
trolled by the degree of resolution obtained by the chromatographic method, and
that the maximum detectable carbon number increased with increase in tempera-
ture (reaching a maximum at
375°C).
These factors all affected the carbon num-
ber distribution and the ratio of total straight to branched chain alkanes. Inde-
pendent molecular sieve analysis showed that the experimental blend contained
approximately
25%
branched chain alkanes. This was found to be similar to the
horizontal baseline result, but very different to the valley-valley and skim base-
line results. The work was repeated on a
15
m
X
0.25
mm i.d., 0.1 pm
DB-1
col-
umn (J&W) programmed to
370°C.
The wax blend (containing
5%
heptadecane
References
pp,
9&93
76
Chapter
3
internal standard) was injected into a PTV, which was programmed to
390°C.
Electronic background subtraction was used together with the horizontal base-
line integration method, which gave total branched chain alkanes
of
30%.
From
this work, the horizontal baseline method appears to give the most accurate re-
s
u
It
s
.
Michalczyk
[67]
published a paper on the quantitative analysis
of
paraffin
waxes, using cool on-column injection onto a
15
m
X
25
m id.,
0.1
pm
OV-101
column. The
GC
oven maximum temperature was
330"C,
whereas the detector
temperature was set at
300°C.
This leads to condensation of the higher alkanes
eluting from the column and distorts chromatographic results.
A
hexadecane in-
ternal standard was used and the background was subtracted electronically. He
gave detailed proposals for two methods of area measurement (see Fig.
3.5).
(a)
For the first method, a horizontal line
is
drawn from just before the first
peak
to
the other side of the chromatogram, forming the baseline. The
main alkane peaks are identified, then perpendicular lines are dropped
from the troughs either side of the peaks to the baseline to enclose the n-
alkanes. The areas between the n-alkane peaks were designated as iso-
alkanes and cycloalkanes, taking into account that the iso-alkanes
of
a
given carbon number elute from the column before the n-alkanes.
(b)
For
the second method, the n-alkanes are integrated as if they were sitting
on
top
of
an unresolved hump, above a horizontal baseline. The unre-
solved area of the hump underneath each of the n-alkanes
is
added to the
total iso-alkane area
for
the next higher carbon number.
Fig.
3.5.
Carbon number distribution integration methods.
The chromatographic analysis
of
refined and synthetic waxes
77
Method (b) was preferred by Michalczyk on the grounds that the total
branched chain alkane figures for this type of integration were (for the paraffin
waxes analysed) more comparable to those obtained by the urea adduct and the
molecular sieve methods for separating branched chain alkanes
[68,69].
How-
ever, the second, higher melting paraffin wax gave anomalous results. There are
problems with obtaining accurate results with the urea adduct method. It has also
been shown that the molecular sieve method can be unreliable, due to variations
in the amount of branched chain alkanes separated by different batches of mo-
lecular sieve
[70].
Therefore, these alternative separation methods are dubious
criteria for justifying an integration method.
The ASTM
D.02
Committee has recently produced a draft method for the
capillary GC separation of petroleum waxes
[71].
This method uses internal
standardization (by adding n-hexadecane) with cyclohexane as solvent.
A
variety
of column types and operating conditions are proposed for separating petroleum
waxes. Interlaboratory tests have been carried out to examine the repeatability
and reproducibility of the method. However, problems have been highlighted
and the method
is
under review.
The development of sophisticated integrators and PC software capable of
storing, reintegrating and redrawing chromatographic data, has enabled the car-
bon number distribution calculation to be more carefully examined. It is now
possible to rapidly collect peak data, subtract the column bleed baseline, exam-
ine the integration points in detail, and alter the integration conditions to obtain
accurate quantitative measurements. As wax chromatograms are complex with
many peaks, it is important to collect data at a rapid rate
(10-20
data points per
peak)
[72].
The column bleed baseline increases rapidly above
300°C.
It needs to
be stored as raw data
[18]
or as an algorithm
[73]
so
that it can be subtracted
from the wax chromatogram to give the true sample baseline. Then the peak de-
tection parameters should be set to accommodate the different peak types (large
n- and small iso-alkanes, plus changes in peak widths over the whole chroma-
togram)
[74].
Finally, the most contentious issue, the method of measuring the
peak areas (peak integration mode) has to be chosen.
With most paraffin waxes and for synthetic waxes, it is possible to set the
chromatographic conditions such that all the n-alkanes are resolved to a horizon-
tal baseline. Under these circumstances the horizontal baseline, baseline-
baseline, or the Michalczyk integration method will all give the same carbon
number distribution results. However, with intermediate and microcrystalline
waxes (or blends of them with paraffin waxes), the n-alkane peak resolution de-
creases and the different integration methods give divergent results. This is par-
ticularly true at the high molecular weight end of the distribution where it can be
difficult to distinguish the n-alkane peak areas from the iso-alkanes. In this
situation (see Fig.
3.9,
it
is important that the position of the integration baseline
References pp.
90-93
78
Chapter
3
and the integration points are clearly marked to establish the exact boundaries of
each peak area measurement. The question then arises as to what is the most ac-
curate way of measuring the area
of
a
major peak when it is overlapped by minor
peaks.
This has been recently addressed in detail by Dyson
[66].
For
most wax
chromatograms, the n-alkanes are the major symmetrical peaks and the appended
iso-alkanes are much smaller, narrower asymmetric peaks. For these conditions,
the error of measuring the area of the smaller peak by dropping a valley perpen-
dicular increases as it becomes more of a shoulder on the side of the n-alkane
peak. However, if the integration software
is
written to construct a curving tan-
gential skim, then the error is much reduced. Where the overlapping peaks are
symmetrically shaped, and
of
the same height and width, then a perpendicular
line can be dropped from the valley to the baseline to enclose the areas. The
valley should be no more than
5%
of the peak height. This condition can occur at
the end of the chromatogram. From recent work carried out
by
the author
[75],
it
appears that
a
combination
of
perpendicular and curved skim lines to a horizon-
tal baseline does accurately demarcate each n-alkane from the group of iso-
alkanes of the same carbon number (see Fig.
33~)).
This is providing that the
chromatographic conditions are adjusted overall to obtain the best n-alkane peak
symmetry and resolution to baseline.
After
identifying the peaks with alkane standards and measuring the peak ar-
eas, each group of iso-alkanes of the same carbon number needs to be totalled
and reported separately from the n-alkanes. If this is carried out manually using
a
calculator, it takes
a
long time, but a PC with the requisite software can save
many hours of laborious work. It
is
possible to convert the binary data output
from a
GC
(or
from an
A/D
converter connected to the GC) into ASCII text files
on
a
PC.
These files can then be imported into a spreadsheet
[76]
(e.g.
Lotus
1-2-
3,
QuattroPro, Microsoft Excel) and made into a table of n-alkane, iso-alkane
and total alkane results. From this table, a carbon number distribution graph can
easily be constructed (see Fig.
3.6).
This procedure can be automated using a
spreadsheet macro
or
a specially written program, which makes the whole proc-
ess
enjoyable! However, it is always essential with this complex type of analysis
to check the results against the chromatogram, particularly
for
mistakes in cali-
bration
or
calculation.
3.3
SUPERCRITICAL FLUID CHROMATOGRAPHY
OF
WAXES
The subject of supercritical fluid chromatography
(SFC)
in the petroleum in-
dustry
is
covered in another chapter and its use for high molecular weight petro-
leum products has been covered previously
[77].
But for waxes, the importance
The chromatographic analysis
of
rejhed and synthetic waxes
79
4.5
4
3.5
3
0
2.5
5
$2
1.5
1
0.5
0
_--
CARBON NUMBER
Fig. 3.6.
Carbon number distribution
of
a
rubber
wax.
and the difficulties of this type of chromatography deserve further comment. In
the mid-l980s, SFC showed great promise as a possible means of resolving high
molecular weight waxes into individual components, because it appeared not to
be restricted by sample volatility. It was thought that SFC would act as a bridge
between the high resolution capillary GC separation of paraffin waxes and the
inadequate resolution of synthetic waxes by size exclusion chromatography.
However, the development
of
HTGLC in the late 1980s has paralleled SFC ad-
vances and has undervalued the use of the latter for wax separation.
For wax analysis,
SFC
is a more complex method than HTGLC. To obtain an
even resolution over the whole carbon number distribution, a combination of
pressure or density, and temperature programming is often required. Using mi-
crobore columns it is possible to obtain a resolution somewhere between packed
column and capillary GC.
Sample injection is more complicated
than
for HTGLC, as a much higher
concentration of wax is required for the split injection system used. This initially
led to difficulties with solubility in the mobile phase and biased quantitative re-
sults. These problems have been largely overcome by the use of heated loop in-
jectors with rapidly rotating valves, which give a minute volume on-column in-
jection. Outlet restrictors have also been improved to give much more accurate
quantitative analysis. The column outlet is normally coupled to an FID, but there
is no reason why evaporative light scattering detection could not be used
[78].
References pp.
90-93
80
Chapter
3
Although SFC analysis does not achieve the resolution of HTGLC, its value lies
in being an alternative method giving another insight into the separation
of
high
molecular weight waxes.
The initial investigations into the separation
of
paraffin waxes were mainly
carried out using 50pm diameter fused silica columns varying in length up to
20
m
[79,80].
The preferred supercritical fluid was carbon dioxide (combined
with pressure
or
density programming), normally combined with FID. Split in-
jection was usually used at ambient temperature, which produced
a
tailing off of
the response factors for alkanes above
C30.
The later introduction
of
heated in-
jectors minimized this effect. Another common problem was the production of
electronic spikes, which is thought to be due to involatile solute particles enter-
ing the detector
1791.
It was also found difficult to obtain an even, high resolu-
tion over the whole carbon number distribution. These last
two
effects can
clearly be seen
in
Fig.
3.7.
Some very advanced packed capillary column SFC research has been carried
out by Hirata and Nakata [81]. Using a small heated solid injector at
120°C
and
pressure programming, they successfully separated Polywax
500
to approxi-
mately
C80
and Polywax
655
to
C94.
Their excellent work in separating syn-
thetic waxes was not improved until the late
1980s [82].
A typical separation of
Polywax
655
is shown in Fig.
3.8.
Hawthorn and Miller managed to separate
a Fischer-Tropsch wax to above
CIOO,
but only resolved the n-alkane peaks
[83].
It has recently been shown that
it
is
possible to separate Polywax
1000
up
I
I
.
.J
0
10
20
Fig.
3.7.
Capillary
SFC
analysis
of
a rubber
wax
blend.
Column,
10
m
x
50pm
DB-5; oven
temp.,
140°C;
mobile phase,
carbon
dioxide;
pump,
Brownlee microgradient
system;
injector,
Valco
C14W
l0Oul
loop
at
55°C;
FID detector,
350°C.
The chromatographic analysis ofrefined and synthetic waxes
81
0 10
20
30
40
Fig.
3.8.
Capillary SFC analysis of
Polywax
655.
Chromatographic conditions
as
before except:
elution conditions
2000-6000
psi in
25
min; total
run
time,
40
min.
to C130 using asymptotic density programming
of
carbon dioxide over
70
min
Advances in equipment and techniques have proved that it is possible to ob-
tain high resolution of alkanes to above
C44
[85], with an even alkane response
factor to C60 [62]. However, it has not been possible to match the resolving
power of HTGLC to produce accurate carbon number distributions of both n-
and iso-alkanes for microcrystalline waxes. Several comparisons of simulated
distillation by HTGLC and
SFC
have been made (one for microcrystalline wax
[77] and one for Polywax
750
[62]). From these papers it can be concluded that,
by using a heated timed split injector, it is possible to obtain virtually the same
molecular weight distribution curve as by HTGLC.
1841.
3.4
SIZE EXCLUSION CHROMATOGRAPHY
Size exclusion chromatography
(SEC),
also
known
as gel permeation chroma-
tography (GPC), has been used for nearly
30
years to separate polymers accord-
ing to molecular size. The separating column (usually made of stainless steel) is
tightly packed with particles (up to 20pm diameter) containing pores
of
a con-
trolled size. A mobile phase is chosen that dissolves the sample, but is compati-
References pp.
90-93