Full text of DOI:10.1093/chromsci/41.10.535

Journal of Chromatographic Science, Vol. 41, November/December 2003
GC: A Review of the State-of-the-Art Column
Technologies for the Determination of ppm to ppb
Levels of Oxygenated, Sulfur, and Hydrocarbon
Impurities in C1–C5 Hydrocarbon Streams
J. de Zeeuw*, C. Duvekot, J. Peene, P. Dijkwel, and P. Heijnsdijk
Varian, Inc., Middelburg, the Netherlands
cial selectivity and inertness is required. Aside from the more
common polar liquid phases, it is shown that using a unique CP-
Abstract
Lowox adsorbent multilayer column technology, oxygenates can
The determination of ppm to ppb levels of sulfur, oxygen-
be selectively retained and quantitated at ppm levels. Using con-
centration techniques such as precolumn and stack injection, it
is possible to determine ppb levels of oxygenated impurities in
hydrocarbon streams with high accuracy. This paper summa-
rizes and reviews the main approaches used for trace analysis of
the three component types.
containing, and certain reactive hydrocarbons in bulk hydrocarbon
feedstocks is important in the petroleum and petrochemical
industry to minimize catalytic deactivation and improve product
quality. Gas chromatography, coupled with selective or ultratrace
universal detection (or both), is ideal in most cases for such
analysis. However, to enhance selectivity and quantitation at the
trace levels, optimized stationary phases are required. These phases
are usually of the adsorbent type. This paper summarizes the
performance of several state-of-the-art phases for the analyses of
trace key hydrocarbons, sulfur, and oxygenated components.
Experimental
Hydrocarbon impurities in hydrocarbon streams
In hydrocarbon feedstock streams, many types of hydrocar-
bons can be present. The presence or absence of hydrocarbon
impurities provides information on final product quality and
value or the applicability of feedstock streams/intermediate prod-
ucts to a petrochemical process. Because of its low detection
limits, stability, and ruggedness, flame ionization detection
(FID) is used most widely. Depending on the hydrocarbon impu-
rity requiring analysis, certain column technologies can be used.
Because these compounds are volatile, adsorption-type columns
find wide application in the industry.
The most universally applicable phase for light hydrocarbon
separation is alumina Al2O3. The latter phase was one of the first
fused-silica porous-layer open-tubular (PLOT) columns devel-
oped, and it has found a wide application in petrochemical anal-
yses, especially for light C1–C5 hydrocarbons, for which this
phase exhibits unique performance.
Introduction
The determination of trace impurities by gas chromatography
(GC) is a very common analysis in the petrochemical laboratory.
Usually, these impurities, which can have a detrimental effect on
catalytic deactivation or product quality (or both), consist of
three types: hydrocarbons and sulfur- and oxygen-containing
organo-molecules. Other impurities such as nitrogen- and
heteroatoms- (arsenic, silicon, etc.) containing components may
also have deleterious effects, but these are not addressed in this
paper.
The analysis of trace hydrocarbons can be accomplished very
effectively using selective adsorbents such as silica, alumina, and
carbon. However, the analyses of trace sulfur compounds and
oxygenates require a higher degree of inertness. For sulfur-con-
taining compounds, a polydimethyl siloxane (PDMS) stationary
phase for generic application or a silica-based stationary phase
for special applications may be adequate.
Hydrocarbon separation using alumina adsorbents
Alumina adsorbents in capillary columns were introduced in
1963 (1) and were commercialized in fused-silica capillary
columns in 1981 (2). The alumina adsorbent has a very high
Oxygenated compounds are highly polar, and additional spe-
* Author to whom correspondence should be addressed: email jaap.de.zeeuw@varianinc.com.
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.
535

Journal of Chromatographic Science, Vol. 41, November/December 2003
activity and retains components as light as ethane. Alumina
requires deactivation to make its highly active surface a viable
stationary phase in GC.
For the majority of applications, the KCl-deactivated alumina
can be used. This column, which now is also recommended in
several methods standardized by the American Standards Testing
Materials, will separate the majority of the C1–C4 hydrocarbons
and is ideal for impurity analysis (Figure 2).
Deactivation can be accomplished in many ways; however, the
most practical and reproducible approach is the deactivation
with inorganic salts (3,4). A very popular deactivation procedure
is the addition of KCl to the alumina that results in a general
nonpolar and less active alumina surface, as demonstrated by the
elution of acetylene ahead of butane (Figure 1). The alumina sur-
face can also be made more polar than KCL deactivation by deac-
tivating with sodium sulfate. Using sodium sulfate, the resulting
alumina layer will elute acetylene after the butane peak, indi-
cating a higher polarity of the phase. In addition, methyl-acety-
lene will be retained longer on a sodium sulfate-deactivated
surface and will elute after 1,3-butadiene.
The selectivity of alumina for hydrocarbons is very high. All of
the C1–C4 hydrocarbons can be separated baseline. The resolu-
tion between the different hydrocarbons is sufficient to deter-
mine many trace C1–C4 hydrocarbons in feedstock streams
containing any C1–C4 as the major component hydrocarbon
(Figures 2 and 3). As a result, the alumina phase is one of the
most widely used columns in the petrochemical industry for the
analyses of hydrocarbon impurities in ethylene, propylene, and
butylene feedstock streams.
In certain applications, greater column resolution may be
required, particularly for cases in which the large feedstock hydro-
carbon matrix component(s) may elute close to the impurities of
interest and interfere with the analysis. In such cases, the major
matrix component will appear as an “overloaded” peak. Under these
conditions, the maximum resolution for butylene isomers is
obtained by using a more active Na2SO4-deactivated alumina
column. For this application, the Na2SO4-deactivated alumina is
the optimum phase for impurity analysis, as the resolution between
cis-2-butene and 1-butene is the greatest. Figure 3 shows an impu-
rities analysis in 99.99% 1-butene. In the presence of the large 1-
butene matrix peak, it is still possible to determine the other C4
isomers. For an effective separation, the injection volume must be
kept as small as possible. The large concentration of the “main” or
matrix component, if not controlled, will influence the peak shape
of the trace components that elute close to the main peak.
Characteristics of alumina phase
Capillary columns prepared with alumina phases are very
rugged. Although alumina has unique separation characteris-
tics, it also has some limitations that require attention. The
activity of the adsorbent will adsorb any water, carbon dioxide, or
other polar impurity in the sample or carrier gas stream. The
adsorbed water will deactivate the surface, and, as a conse-
quence, the column will have lower retention of the analytes (2).
The water can be removed simply by heating the alumina capil-
lary column to 200°C for 15 min. The water will desorb, and the
column will be regenerated. When an analysis requires
isothermal operating conditions, a polar precolumn that retains
the water may be used. A column coated with 1–2-µm film thick-
ness of polyethylene glycol is very effective, as the C1–C6 hydro-
carbons will elute from the precolumn and the water will be
retained on the precolumn. The water may be removed from the
KCI
Na2SO4
Time (min)
Figure 1. Selectivity of KCl and Na2SO4-deactivated Al2O3 surfaces: the
Na2SO4-deactivated surface has more retention for polar hydrocarbons.
50 m × 0.32 mm Al2O3–Na2SO4 5 µm
Temperature: 110°C
Time (min)
Time (min)
15
Figure 2. Impurities in 1,3-butadiene on Al2O3–KCl; methylacetylene elutes
Figure 3. Impurity analysis in 99.99% 1 butene; using Na2SO4-deactivated
in front of 1,3-butadiene.
alumina will give the best separation.
536

Journal of Chromatographic Science, Vol. 41, November/December 2003
precolumn by simple backflush or a vent-switching flow config-
uration (5). For more deleterious polar contaminants, the alu-
mina column regeneration may take longer, but is possible. It
was found that sulfur impurities present up to 2000 ppm did not
seriously affect the stationary phase or the retention times of the
trace impurities on alumina. Oxygen in the carrier gas will not
harm the column. The maximum temperature of alumina
phases should not exceed 200°C because above 200°C, recrystal-
lization of the deactivation crystals will occur, causing irre-
versible changes in selectivity (4). Alumina layers coated in
Ultimetal tubing (Varian, Middleburg, the Netherlands) are very
stable and find wide application in stressful environments (7).
In summary, the aluminum oxide phase has now been used
by many laboratories in petrochemical and environmental
analysis and is recognized to be the most effective phase
capable of resolving most of the C1–C5 hydrocarbons with max-
imum separation efficiency.
lowering the GC oven temperature programs, and (c) using
shorter columns. All of these actions will improve the analysis. In
addition, a new alumina phase column called Select Al2O3–
methylacetylenepropadiene (MAPD) (Varian) was developed that
has an improved deactivated surface that exhibits a much lower
retention for the hydrocarbons. The lower retention characteris-
tics of this improved alumina phase allow the use of lower
column temperatures and, thus, a higher response for the polar
hydrocarbons. The results obtained with the improved phase are
shown in Figure 4. The response for propadiene is much higher,
allowing lower concentrations to be determined. This improved
phase will yield a greater response for acetylene and methyl
acetylene.
Acetylene in ethylene
The determination of acetylene in ethylene is an important
analysis in the petrochemical industry. To separate the acetylene
from ethylene adsorbents exhibits ideal selectivity. Aluminum
oxide may be used; however, for any trace impurity analysis, it
is important that the impurity elute significantly in front of
the main component or that it elute significantly after the
main component. In both cases, low concentration levels can
be determined.
Trace analysis of methylacetylene, acetylene, and propadiene
Although alumina is the ideal adsorbent for most hydrocarbon
analysis, it has some disadvantages because of its high surface
activity, which results in an irreversible adsorption for the more
“polar” hydrocarbons. These “polar” hydrocarbon compounds
are typically methyl acetylene, acetylene, and propadiene.
Although these compounds can be quantitated using alumina,
the response of the detector for such components is lower than
expected, based on their theoretical carbon number response
using the FID. However, with frequent calibration, it is possible
to obtain acceptable quantitation of these compounds. To
improve the analysis using alumina for the determination of
these compounds, a study was undertaken (8) to better under-
stand the elution characteristics of these compounds as a func-
tion of column operating conditions. It was determined that the
cause for the low response was an irreversible adsorption that
can occur at higher column temperatures, which can be
improved by reducing the elution temperatures. The lowering of
the elution temperatures can be obtained in several ways: (a)
with existing columns using higher carrier gas flow rates, (b)
Methane
Ethane
(200 ppm)
Acetylene
50-m × 0.53-mm
fused silica
(3 ppm)
CP-CarbonBOND
Figure 5. Analysis of ppm levels of acetylene on a CP-CarboBOND column.
Acetylene elutes before ethylene. Symmetrical peak allows sub-ppm level
determination (courtesy of J. Luong, Dow Chemical, Canada).
Column: 30-m × 0.53-mm Varian Select Al2O3–MAPD
Oven: 40°C, 5 min – 10°C/min ꢀ 180°C, 20°C/min ꢀ 200°C.
Carrier gas: He, 4 psig – 4 min – 0.5 psig/min – 11 psig – 2 min
Column: 30-m × 0.32-mm CP-SilicaPLOT
Carrier gas: Helium, 134 kPa
High response for propadiene
Cyclopropane (1 ppm)
1
2
3
4
5
6
7
8
9
Methane
Ethane
1 Methane
2 Ethane
3 Ethylene
4 Acetylene
5 Propane
6 Cyclo Propane
7 Propylene
8 C4-hydrocarbons
Ethylene
N-butane
Propadiene
1-Butene
Iso-butene
1,2-Butadiene
1,3-Butadiene
10 Ethyl acetylene
Time (min)
Time (min)
22
Figure 4. Elution of polar hydrocarbons on Select-Al2O3–MAPD. Note high
response for acetylene and propadiene (courtesy of J. Luong, Dow
Chemical, Canada).
Figure 6. Impurity analysis of ppm-level cyclopropane in propylene;
silicaPLOT makes the cyclopropane elute in front of the propylene.
537

Journal of Chromatographic Science, Vol. 41, November/December 2003
For the determination of trace acetylene in ethylene, two
materials are of greater interest than alumina, silica and carbon.
Using silica, the acetylene will elute after the ethylene peak (9).
Separation is acceptable, but the acetylene elutes on the tail of the
ethylene. This may me acceptable for the analysis of high ppm
levels, but for trace analysis at lower concentrations, an improved
separation is required. A highly unique selectivity is obtained with
the carbon deposited adsorbents (10). These types of adsorbent
materials have been used extensively for a long time in packed
columns, under the name Carbosieve, as coated carbon materials.
Several attempts have been made to commercialize capillary
columns coated with Carbosieve materials, however, not with great
success. Presently, two carbon types have found a practical applica-
tion, the Carboxene and CarboBOND type of materials. Both mate-
rials exhibit comparable characteristics, as they are highly nonpolar
and elute acetylene in front of ethylene. The main limitation for the
application of the carbon-coated PLOT columns is the limited tem-
perature range. Carbon-deposited layers become very active upon
thermal exposure. Usually, application of these materials is limited
to approximately 200ºC, but up to this temperature, unique separa-
tions can be performed.
lene. Because the acetylene elutes before the main ethylene peak,
it can be determined to low concentration levels. Single-column
analysis down to 500 ppb acetylene should be attainable.
Cyclopropane in propylene
In certain chemical processes, there is an interest in the deter-
mination of cyclopropane. This component is very volatile, and
its determination requires special columns. Alumina adsorbents
separate the cyclopropane, but it cannot be determined at ppm
CP SilicaPLOT 30 m × 0 32 mm
40°C 5 min ꢀ 200°C 10°C/min
A
MS-TIC
Time (min)
7
Figure 5 shows the analysis of ppm levels of acetylene in ethy-
CP SilicaPLOT 30 m × 0 32 mm, df = 4 µm
30°C, 8 min ꢀ 250°C 10°C/min
B
CH3SH
Sample: 1 ppm via 6 port GSV; PFPD detection
H2S
COS
3.0
4.5
6.0
7.5
9.0
10.5
12
Time (min)
20
Time (min)
Figure 7. Separation of most common sulfur compounds on thick-film PDMS
(CP-Sil 5 CB) using SCD detection (courtesy of J. Luong, Dow Chemical,
Canada).
Figure 9. (A) Separation of COS and propylene on a silica PLOT column;
COS is well resolved from propylene allowing ppb-level detection with
selective detectors. (B) Determination of trace COS in propylene matrix; no
quenching is possible, as COS is well separated from propylene.
30 m × 0.32 mm CP-SilicaPLOT df = µm
40°C, 5 min ꢀ 200°C, 10°C/min
1000 ppm in air; split injection: MS detection
10 m × 0.53 mm Ultimetal Simdis CB, 0.5 µm
35°C ꢀ 380°C, 20°C/min: Helium, 40 kPa, SCD
Sample: Sulfur in C12–C60 wax
H2S
CO2
Time (min)
25
Time (min)
20
Figure 8. Separation of sulfur compounds on a silica PLOT column; COS and
H2S elute at high temperature and are well resolved. Also, SO2 shows good
peak shape.
Figure 10. Separation of sulfur compounds in a C12–C60 wax sample using an
Ultimetal Simdis column (courtesy of J. Freelin, Shell Westhollow, Houston, TX).
538

Journal of Chromatographic Science, Vol. 41, November/December 2003
levels. An Al2O3–KCl column coupled with a 50-m × 0.32-mm
CP-Sil 5 CB (Varian) with a 5-µm film was used successfully for
such analysis (11). However, the analysis time was long and the
cost of the required columns was high because it required using
a 100-m coupled-column configuration. The same analysis can
be performed using a single silica adsorbent, as shown in Figure
6. Because the cyclopropane elutes in front of the propylene, it
can be determined at very low concentration levels. In this appli-
cation, the oven was heated from 50ºC (5 min) with 5ºC/min up
to 100ºC, followed by a 20ºC/min program to 225ºC. The sample
was introduced with a Valco sampling valve (VICI, Houston, TX)
with a 250-µL sample volume into the splitter injector at a split
ratio of 10:1. Even at the low concentration levels, the cyclo-
propane eluted in front of the propylene peak and could be quan-
titated accurately.
High-ppm up to percent-level determinations can be accom-
plished with thermal conductivity detection (TCD). With
microchip-TCD design, sulfur compounds can be determined to
10-ppm levels (see following section on sulfur in natural gas).
Using micromachined technology for injection and detection
devices, it is possible to have very short analysis times (< 200 s)
with high sensitivity. Because detection is performed with a
microchip-TCD (universal thermal conductivity detector), the
selectivity of the column becomes very important in separating
the sulfur compounds from the matrix.
The main disadvantage of the TCD is its limited sensitivity and
its universal response. In addition, long-term exposure to reac-
tive sulfur compounds (e.g., SO2) limits its lifetime.
Sulfur selective detection is well developed and is used very
broadly. Sulfur determination using chemiluminescence detec-
tion (SCD) allows low levels of sulfur compounds to be deter-
mined. Because of its equimolar response, the SCD has high
selectivity and low detection limits at ppb levels. Alternatively, a
pulsed flame photometric detector (PFPD) may be used (12). The
PFPD also allows ppb levels to be determined. However, the
response of this detector is quadratic, which can be accounted
for with the data interpretation. Although these selective detec-
tors have nearly six decades of selectivity, at ppb levels, the sulfur
signal “quenching” effect from hydrocarbon background may
yield a nonlinear response or interference from coeluting large
amounts of hydrocarbon components (or both). To minimize
interferences from the hydrocarbons, a column phase must be
used that separates the sulfur components from the interfering
background. Generally, as more selective detection systems are
used, the selectivity of the column becomes less important; how-
ever, for trace sulfur compounds analysis, a good separation from
the large hydrocarbon matrix components almost always will
lead to a more accurate analysis.
Sulfur impurities in hydrocarbon streams
Sulfur compounds exist in light feedstocks in a broad boiling
point range varying from the very volatile sulfur gases to high-
molecular-weight thiophenes. A number of sulfur compounds
are very reactive, requiring an inert GC system for trace sulfur
determination. The separation of sulfur compounds can be
accomplished very well with GC, as these materials are relatively
nonpolar and can be separated on different types of stationary
phases. The main stationary phases used for separation of sulfur
compounds are nonpolar phases based on polydimethylsiloxanes
or PLOT columns coated with porous polymers or silica. The
PLOT columns are mainly used for volatile sulfur compounds
like carbonyl sulfide, hydrogen sulfide, and methylmercaptan,
which are also the most reactive.
Detection of sulfur compounds
For the detection of sulfur compounds, several detectors can
be used depending on the concentration levels to be determined.
Sulfur SimDis
Hydrogenated naphtha
Column
: 10 m × 0,53 mm Ultimetal CP-Sil 5 CB. df = 5 µm
Detection : SCID Antek
Method: D-37 10-S
Carrier
Oven
: He Const Flow Rate 15 cc/min
: 35 (1 min ) – 15°C/min – 200°C (2 min)
Total S: 240 ppm
Intel PTV : Hold 0 Rate 15°C/min Final 200°C
ppm Sulfur distribution by boiling point
Boiling point (F)
Figure 11. Total sulfur analysis using an Ultimetal Simdis column, 10 m × 0.53 mm with 5 µm, detection SCD (courtesy of J. Grills, Envantage).
539

Journal of Chromatographic Science, Vol. 41, November/December 2003
Sulfur analysis at ppm–ppb levels
in high-purity propylene using a selective detector (PFPD or SCD).
In contrast, a similar analysis using a nonpolar PDMS liquid phase
(CP-Sil 5 CB or equivalent) caused COS to elute at the same reten-
tion time as the propylene, which would cause sulfur signal
quenching. Figure 9 shows the elution of COS, as well as propy-
lene, on a silica adsorbent. Propylene is well separated, allowing
good response at low levels for the selective detectors.
Sulfur compounds need to be determined in a wide variety of
hydrocarbon matrices. Although specific columns are available,
many “generic” analyses have to be performed. For analyses of
general sulfur compounds in up to C7–C8 hydrocarbons, a non-
polar, such as 100% polydimethylsiloxane (PDMS), stationary
phase will generally yield the best results. PDMS can be coated
on deactivated fused-silica (or UltiMetal, or equivalent) surfaces,
allowing ppb levels of sulfur compounds to be quantitated. For
example, Figure 7 shows the separation of the C1–C7 sulfur com-
pounds on a 5-µm CP-Sil 5 CB phase at 30ºC. However, this sep-
aration may be problematic for the analysis of H2S, COS, and SO2
at the specified temperature. Because these compounds, espe-
cially COS and SO2, elute very closely together close to the
column dead volume, the quality of the injection becomes a
major issue. The optimum solution is to use PLOT columns as
described next.
High-boiling sulfur compounds, sulfur simdis,
and total sulfur analysis
To analyze the total sulfur content of hydrocarbon streams, it
is necessary to elute all of the sulfur compounds, including
those with a higher boiling point. Metal capillary columns are
commercially available that possess a high degree of inertness
for sulfur analyses. Figure 10 shows the sulfur impurities in a
C12–C60 wax sample for which elution temperatures up to 380ºC
were required. Using the SCD in combination with thick-film
polydimetylsiloxane stationary phase, it is possible to visualize
the sulfur distribution. Figure 11 shows the fingerprint of a cat-
alytically cracked naphtha sample containing a total of 240 ppm
sulfur. The inert UltiMetal column used allows the higher
boiling volatile compounds to elute.
Another example of using selectivity of the stationary phase is
the analysis of thiophene in benzene with SCD. The analysis
may be performed very well using a polyethylene glycol-type
stationary phase (CP-Wax 52 CB or equivalent) at oven temper-
atures starting at 30ºC with 10ºC/min temperature program-
ming to 125ºC. The thiophene (11 ppb) elutes as
Determination of H2S, SO2, and COS
One option for the determination of H2S, COS, and SO2 is to
use a selective stationary phase based on SIO2 (silica) (10). Silica
PLOT columns have the advantage that they elute not only the
hydrocarbons, but also the sulfur compounds at low concentra-
tion levels. The sulfur compounds are retained and separated at
temperatures as high as 80ºC. This makes the separation of COS
and SO2 quite simple, as shown in Figure 8.
The silica PLOT column is also effective in the analysis of COS
a resolved peak with minimal interference, thus
allowing low-level determination in the large
benzene matrix (Figure 12).
A completely new approach for the determina-
tion of total sulfur was presented by Amirav (16),
as shown in Figure 13. A PFPD detector was
used, whereby the sample was introduced
directly via a short 0.53-mm capillary coated
2
4
6
8
10
12
Figure 12. Analysis of thiophene in benzene. Polyethylene glycol allows low-level determination of
thiophene using selective detection.
with approximately 0.5-µm film of poly-
dimethylsiloxane. The 0.53-mm short column
was coupled with a 1-m × 0.1-mm capillary to
create an acceptable pressure drop to allow con-
trol of the carrier gas flow. The sample was
directly introduced in the PFPD, giving a
response for sulfur. This method generated a
total sulfur value within 100 s and can be used to
approximately 5 ppm. Lower sulfur contents are
difficult to attain because of the response of the
matrix or major coeluting hydrocarbons.
Column: 1 m × 0.53 mm, with restriction
1 m × 0,10 mm, deactivated
Oven:
280°C; Helium 0,1 mL/min
PFPD detection
Sulfur in natural gas
Natural gas is used widely for cooking and
heating applications. One of the problems of nat-
ural gas is that it has no odor. As a result, very
risky situations can occur that can lead to per-
sonal injuries. To avoid safety problems, an
“odorant” is added to natural gas to allow detec-
tion of small gas leaks. Very low concentrations of
sulfur compounds are added as odorant because
the human nose is very sensitive for sulfur com-
100 s
0
20
40
60
80
100
Figure 13. Total sulfur determination using a 1-m × 0.53-mm CP-Sil 5 CB coupled with a 1-m × 0.1-
mm deactivated fused silica (courtesy of Aviv Amirav, University Tel Aviv).
540

Journal of Chromatographic Science, Vol. 41, November/December 2003
pounds at fg levels. Several odorants may be added, such as
methyl mercaptan, tetra hydrothiophene (THT), or butyl mer-
captan (or all three). Consequently, rapid analyses are required for
these sulfur components in natural gas at concentrations
between 10–50 ppm. The TCD is a viable detector for these appli-
cations. For speed of analysis, a narrow-bore column would be
ideal. The introduction of micromachined systems for injection,
as well as detection (TCD), has led to the development of microGC
technologies for very fast analysis of odorants in natural gas
steams. Typically, an analysis with the chip technology takes
approximately 50 s. Because the detector in such systems are TCD
based (universal detection), the selectivity of the column becomes
very important. Special stationary phases have been developed for
tetra-hydro-thiophene (THT) and tert-butylmercaptan (TBM) to
elute these components free from interference. Figures 14A and
B show the separation of THT and TBM obtained using a fast
technology used in the CP4900 micro GC (Varian).
Single-column separation of oxygenated compounds in
hydrocarbon matrix
The most desirable approach to analyze trace level of oxy-
genates would be to inject the sample directly into a single
column capable of resolving the oxygenates of interest with min-
imal interference from hydrocarbons with FID detection. To
achieve this goal, a highly selective and stable stationary phase
has to be used, and the column must have a high resolving
power. Generally, this means long columns of small internal
diameter, which result in relatively long analysis times. If there
are only a few oxygenated compounds to be analyzed, the appli-
cation of a single-column analysis becomes more realistic. The
practical limitation on oxygenate levels that can be determined is
approximately 5 ppm.
There are several stationary phases that are known to be selec-
tive for oxygenated components. Polyethylene glycols (PEGs),
cyanosilicones, and the multilayer (CP-Lowox) phases are all
used. Table I gives a comparison of these phases. It is clear that
the CP-Lowox phase offers major advantages with respect to
selectivity, temperature stability, bleed, trapping effect, stability,
and chemical inertness.
Oxygenated impurities in hydrocarbon streams
Introduction
The determination of low levels of oxygenates may be per-
formed using various technologies. These tech-
niques include selective enrichment with
A
adsorbents, multidimensional GC employing
columns with different selectivity, or selective
detectors such as the oxygen-specific flame ion-
ization detection (O-FID) and atomic emission
detection (AED). Although these techniques
performed adequately, key limitations include
difficulty to maintain and the high cost to imple-
ment.
The determination of ppm and sub-ppm levels
of alcohols and aldehydes in different hydro-
carbon matrices is a very important analysis in
the petrochemical laboratory. The analysis of
these “oxygenates” is important because the
level of the oxygenates directly affects catalyst
poisoning or product quality. Usually, this anal-
Time (s)
ysis is performed on high-polarity liquid phases,
such as tri-cyano-ethoxy-propane (TCEP), or
with the high maintenance O-FID, often with
complex valve switching. With these methods,
the detection limit is approximately 10 ppm. In
1995, a new adsorbent stationary phase material
was introduced, called CP-Lowox (Varian), with
an exceptionally high selectivity for oxygenates
because of its multilayer technology. Methanol
has a retention index higher than 1400 at tem-
peratures > 200ºC, which makes its determina-
tion possible at very low levels in ethylene,
propylene, butene, and C5 streams using stan-
dard FID detection. The CP-Lowox column has
virtually no bleed, even at temperatures of
350ºC. Many common oxygenates can be quanti-
tated at sub-ppm and even at ppb levels. Besides
methanol, a range of oxygenated compounds
can be separated and quantitated such as alco-
hols, ethers, aldehydes, ketones, and esters.
B
CP-SIL 13 CB SPECIAL TBM
Temperature : 40–45°C
Inj. Time
Carrier
: 255 ms
: He
Ret time
: 30–40 s
Time (s)
Figure 14. Fast analysis of odorants in natural gas using the CP4900 micro GC: analysis of THT (A)
and analysis of TBM (B).
541

Journal of Chromatographic Science, Vol. 41, November/December 2003
Analysis of methanol
effective, as the large matrix hydrocarbon peak will mask the
methanol peak. For the latter application, only TCEP and CP-
Lowox will be effective.
In a range of 50 ppb–100 ppm methanol, no liquid phase will
be effective, and the CP-Lowox adsorbent phase is the only phase
that can separate methanol at these low levels. Figure 15 shows
the analysis of methanol in pure propylene starting at 175ºC.
Methanol in C1–C3 stream. Methanol is a compound that is
often present in light hydrocarbon streams. It is added to
natural gas during its transportation to prevent freezing of
piping, etc. Depending on the level of methanol and the
matrix, several stationary phases can be used for the analysis.
In the range of 1–100 ppm methanol in methane and
ethylene, the methanol can be determined very simply using
polar phases such as CP-Wax 52 CB or equivalent, TCEP, and
CP-Lowox. If the same amount of methanol is analyzed in
propylene, the CP-Wax 52 CB, or equivalent, phase will not be
Methanol in C4. The analysis of methanol in butadiene at the
100-ppm level is possible using the polar TCEP stationary phase.
There is, however, a coelution on TCEP of methanol with 4-vinyl
cyclohexene (4-VCH), a component that is often present in the
1,3-butadiene matrix. Although TCEP is a very polar and selec-
tive phase, the methanol determination is not possible. The sep-
aration on the CP-Lowox column is shown in Figure 16. The
4-VCH (peak 2) elutes very quickly and is well resolved from the
methanol and other oxygenated compounds. Also note the excel-
lent peak shape for methanol and acetaldehyde.
Table I. Characteristics for Four Polar Phases Used for
Oxygenes Analysis
PEG
100% Cyano CP-TCEP
CP-Lowox
Polarity
Selectivity
medium
high
250
high
low
225
10
high
medium
140
very high
very high
350
Tmax (°C)
Stack injection
Bleed at Tmax (pA)
Trapping effect
0.53-mm Columns
10
low
yes
15
1
To determine low-level oxygenated compounds using a single
column approach, a new technology was developed that is called
“stack injection” (15). The analysis is conducted by performing
successive injections of the same sample with a sampling valve
while the oven is held at a constant low temperature. Light
hydrocarbons, with very low k’ values on this column, elute
rapidly, whereas oxygenated compounds (higher k’ value) from
low
no
low
no
very high
yes
propylene
10 m × 0.32 mm CP-Lowox
150°C (2 min) ꢀ 200°C, 10°C/min
Direct injection 50 µL
FID detection
130.0
methanol 20 ppm
120.0
110.0
100.0
Time (min)
0
9
0.0
2.0
3.0
4.0
5.0
6.0
RT
Figure 15. Analysis of methanol on propylene with 10-m × 0.53-mm
CP-Lowox.
Figure 17. Stack injection technique on a 10-m × 0.53-mm CP-Lowox: (A) 1
injection at 50°C, followed by temperature program to 200°C and (B) 10
injections at 50°C, followed by temperature program to 200°C. Sensitivity
enhancement of a factor 10. (1) Acetaldehyde, (2) propanal, (3) methanol,
and (4) acetone.
10 m × 0.53 mm CP-Lowox
1.3-Butadiene
175°C (2 min) ꢀ 275°C, 10°C/min
Split via Valco valve, 0.1 µL
4-vinyl
cyclohexene
FID detection
methanol ca 20 ppm
acetaldehyde
Figure 18. Setup of a universal analyzer system for analysis of trace oxy-
genated compounds in gas and liquid hydrocarbon matrices up to C12: (F)
flow controller, (P) pressure controller, (S) solanoid valve, (GSV) gas sampling
valve, (LSV) liquid sampling valve, and (NV) needle valve.
Time (min)
Figure 16. Analysis of methanol in 1,3-butadiene using 10-m × 0.53-mm
CP-Lowox.
542

Journal of Chromatographic Science, Vol. 41, November/December 2003
the sum of all of the injections are trapped and refocussed as a
tight narrow band at the front of the CP-Lowox column.
Enhancement of the sensitivity is proportional to the number of
injections made. As an illustration, a stack of ten injections
delivers a detection limit of 35 ppb (w/w) of methanol in pentane
(Figure 17). The stack approach is applicable for a variety of
polar compounds such as aldehydes, ketones, and alcohols. The
technique is very easy to implement with a high degree of relia-
bility. The concept of “stack injection” can be employed for
other analytical applications when sample enrichment is
required for sensitivity enhancement.
coated with CP-Sil 5 CB. The oxygenated impurities will all
elute together in the same range as the C1–C6 fraction. This
fraction is injected onto the CP-Lowox column, and all heavier
components are backflushed. By injecting a large sample onto
the precolumn, it is possible to determine ppb levels of oxy-
genates in naphtha streams using the described configuration
(Figure 19). Such a system can be automated and is suitable for
liquid, as well as gas, samples.
Injection of pressurized (gas) samples
Very often, oxygenates need to be determined in gaseous sam-
ples. The gas sample can be introduced directly into the CP-
Lowox column using its normal inlet pressure of 10–30 kPa. The
column can also be connected with a restrictor at the inlet,
allowing a higher inlet pressure. The higher inlet pressure will
compress the sample size and improve the injection, as shown in
Figure 20. The peak shape of the major peaks will improve,
which will in turn improve the quantitation of
Oxygenates impurities in C1–C12 streams by
multidimensional GC
The determination of oxygenated impurities in hydrocarbon
streams up to C12 is possible using a setup as shown in Figure
18. Here, the injection is performed on a nonpolar precolumn
impurities at low levels. The restrictor can either
be a 3-m × 0.25-mm, 1-m × 0.15-mm, or 0.5-m ×
0.1-mm fused-silica capillary, as long as it is
properly deactivated.
Conclusion
GC is widely applied for trace impurity analysis
in hydrocarbon streams. By selecting the optimal
column, detectors, and system configuration, it is
possible to determine trace levels of hydrocar-
bons, sulfur compounds, as well as oxygenated
compounds. The applications presented in this
paper have all been implemented in various labo-
ratories and have produced reliable data.
25 min
Figure 19. Example of oxygenated impurities in naphtha as performed with the system shown in
Figure 18; besides methanol, a large range of oxygenated compounds can be determined.
Acknowledgments
The authors acknowledge the contribution of
Frank P. Di Sanzo, ExxonMobil Research and
Engineering, Annadale, NJ in the preparation of
the manuscript.
Without restriction
References
1. J.J. Kirkland. Anal. Chem. 35: 1295–97 (1963).
2. R. de Nijs. Analysis of light hydrocarbons C1–C5
with porous layer open tubular fused silica
columns of aluminium oxide. J. HRC&CC. 4:
612–15 (1981).
With restriction
0.5 m × 0.1 mm methyl deac. f.s.
Inlet pressure: 400 kPa
3. W. Schneider, J.C. Frohe, and H. Brudereck.
Determination of hydrocarbons in the parts per
million range using glass capillary columns
coated with aluminium oxide. J. Chromatogr. 155:
311–27.
4. R. de Nijs and J. de Zeeuw. Aluminum oxide
coated fused silica porous layer open tubular
Figure 20. Using a restriction at the front of the 0.53-mm column allows injection at higher inlet pres-
sure, resulting in improved peak shape.
543

Journal of Chromatographic Science, Vol. 41, November/December 2003
column for gas solid chromatography of C1–C10 hydrocarbons. J.
Chromatogr. 279: 41–48 (1983).
5. D.J. Breeman. Backflush column removes GC sample water.
Hydrocarbon Processing March: 90–91 (1990).
6. J. de Zeeuw and R.C.M. de Nijs. Adsorption chromatography on
PLOT (porous layer open tubular) columns: a new look at the future
of capillary GC. J. Chrom. Sci. 25: 71–83 (1987).
7. H. Heijmans, J. de Zeeuw, J. Buyten, J Peene, and M. Mohnke. The
use of PLOT columns in gas chromatography. Int. Chrom. Lab. 22:
2–6 (1994).
Varian BV, Middelburg, the Netherlands, 1988.
12. E. Atar, S. Cheskis, and A. Amirav. Pulsed flame—a novel concept
for molecular detection. Anal. Chem. 63: 2061–64 (1991).
13. A. Amirav and H. Jing. Simultaneous pulsed flame photometric and
mass spectrometric detection for enhanced pesticide analysis capa-
bilities. J. Chromatogr. A 814: 133–50 (1998).
14. J.de Zeeuw, C. Duvekot, M. Mohnke, D. Estel, and N. Vonk.
Analysis of ppm–ppb levels of oxygenates in C1–C5 hydrocarbons
using a highly selective multilayer PLOT column. American
Laboratory News February: 81–82 (1999).
8. J. Luong, C. Duvekot, and J. de Zeeuw. The Development of a
Highly Sensitive and Stable Al2O3 Column Technology for Reactive
Alkenes and Alkynes. Poster presentation at 27th International
Symposium, Las Vegas, NV, May 2003.
9. M. Mohnke, D. Estel, J.de Zeeuw, and C. Duvekot. A new silica
PLOT column for analyzing volatile compounds. Am. Lab. May:
22–24 (1997).
15. J. Luong, L. Sieben, J. de Zeeuw, and C. Duvekot. Analysis of ppb
Level of Oxygenated Components in Hydrocarbon Matrices Using
Highly Selective Columns, Specialized Injection Technologies and
FID. Proceedings from the 24th International Symposium on
Capillary Separation Technology, Las Vegas, NV, May, 2001.
16. A. Amirav. “Applications of the PFPD”. In Pulsed Flame
Photometric Detector for Gas Chromatography, ref. no. 03-
914625-00. Varian, Walnut Creek, CA, 1998, p. 21.
10. J. de Zeeuw. “Gas solid chromatography”. In Encyclopedia of
Separation Science, 1st ed. Academic Press, San Diego, CA, 2000,
pp. 481–88.
11. Varian BV. Impurities in propylene. Application Note A01267.
Manuscript accepted September 10, 2003.
544