Comprehensive Two-Dimensional Gas Chromatography with Time of Flight Mass Spectrometry Applied to Analysis of Fischer-Tropsch Synthesis Products Obtained with and without Carbon Dioxide Addition to Feed Gas

Worldwide demand for clean, reliable and environmental friendly energy sources has never been greater. Based on that scenario, Gas to Liquids (GTL) solutions, a group of technologies that can create liquid hydrocarbon fuels from natural gas, has become more and more attractive as a way to commercialize stranded gas reserves. In this context, stranded reserves mean any natural gas field which the infrastructure of transportation to the market is not feasible for technical or economic reasons. Also, GTL could be a possible solution to limit natural gas flaring or venting. An important part of GTL is the Fischer-Tropsch (FT) synthesis, which is a catalytic reaction wherein a mixture of carbon monoxide (CO) and hydrogen (H2) called syngas (from methane reforming process) are converted into hydrocarbons of wide range of molecular weights often called synthetic crude (syncrude). The Fischer Tropsch synthesis is so far one of the best ways to convert natural gas into liquid fuels. Conversion of syngas through the FT process leads to a distribution of products essentially consisting of n-alkanes together with smaller percentages of branched alkanes, alcohols, alkenes. The main reactions involved in FT synthesis are shown in Figure 1.


Introduction
Worldwide demand for clean, reliable and environmental friendly energy sources has never been greater.Based on that scenario, Gas to Liquids (GTL) solutions, a group of technologies that can create liquid hydrocarbon fuels from natural gas, has become more and more attractive as a way to commercialize stranded gas reserves.In this context, stranded reserves mean any natural gas field which the infrastructure of transportation to the market is not feasible for technical or economic reasons.Also, GTL could be a possible solution to limit natural gas flaring or venting.An important part of GTL is the Fischer-Tropsch (FT) synthesis, which is a catalytic reaction wherein a mixture of carbon monoxide (CO) and hydrogen (H 2 ) called syngas (from methane reforming process) are converted into hydrocarbons of wide range of molecular weights often called synthetic crude (syncrude).[3] Conversion of syngas through the FT process leads to a distribution of products essentially consisting of n-alkanes together with smaller percentages of branched alkanes, alcohols, alkenes.The main reactions involved in FT synthesis are shown in Figure 1.
The FT reactions occur under catalytic conditions, Fe and Co are the most used catalysts, although Ru, Ni and other transition metals have also been tested. 4Although the FT has been known since the early 20th century, its reaction mechanism is not completely elucidated yet.Two main proposals are dominant: the oxygenated mechanism 5,6 and the surface carbide mechanism. 7,8Investigations about the FT mechanism are made even nowadays.Recently published evidences suggest that FT reaction, under iron catalyst, occurs by the surface carbide mechanism, 9 while the oxygenated mechanism prevails when the reaction occurs under cobalt catalyst. 10he FT heavy products from the low temperature FT process can be upgraded to obtain high quality products such as diesel, jet fuel, petrochemical naphtha, lubricants and food grade waxes.Such cuts are considered sulfur and nitrogen free and the level of aromatic compounds is extremely low (less than 1 ppm).][13] Besides the mentioned benefits, GTL technology could also be used as a means to mitigate CO 2 emission via carbon capture through CO 2 chemical conversion into liquid fuels. 14,15The CO 2 source can be obtained from the plant emission (combustion process, side reforming reaction and FT reaction), from external sources like CO 2 rich natural gas, different types of anthropogenic processes, and even from the ocean. 16ormally the products formed are dispersed in liquid and solid phases.In addition, many isomers are produced, which makes FT products complex samples.The analysis of FT samples is crucial for process monitoring and ensures that the wanted products are obtained.However, the complexity of the FT samples requires analytical techniques with high separation power and comprehensive two-dimensional gas chromatography (GC × GC) is a good option to analyze them.
GC × GC was earlier described by Lui and Philips, 17 the two-dimensional system consisting in two columns with different stationary phases which are connected in tandem and all compounds that elute from the first column are injected in the second one.In GC × GC the compounds are submitted to two separation mechanisms which distributes the compounds in an orthogonal plane, increasing the information about the samples.][20] GC × GC can be coupled to several kinds of detectors, 21 but the most powerful combination is GC × GC coupled to time of flight mass spectrometry (GC × GC-TOFMS) because of its high acquisition rate (up to 500 spectra s -1 ). 22n this work, GC × GC-TOFMS was used to the analysis of Fisher-Tropsch-oil products to evaluate the CO 2 effect on product composition.

Sample description
The FT products were obtained using Co/Al 2 O 3 catalyst in a fixed bed reactor operating at T = 220 ºC, P = 25 bar and H 2 /CO ratio of 1.9.Two Fischer-Tropsch streams, named as FT1 and FT2, were generated and analyzed in this study; FT1 was obtained from synthetic syngas (CO + H 2 mixture from gas bottles) while to produce FT2, the synthetic syngas was blended with 10% of CO 2 .The samples were fractionated in liquid and solid phases employing separators operating under controlled temperature and pressure.The solid fraction is obtained under high pressure and high temperature and liquid fraction is obtained under high pressure and low temperature, both fractions are obtained free of water.The samples were named as follows: FT1 liquid phase (FT1-L), FT2 liquid phase (FT2-L), FT1 solid phase (FT1-S) and FT1 solid phase (FT2-S).Liquid samples were injected directly, while the solid samples were injected as a 100 ng µL -1 solution in dichloromethane.
GC conditions for liquid samples: 300:1 split mode injection of 1 µL at 250 °C.Helium was used as carrier gas at a constant flow rate of 1.0 mL min -1 .The primary oven temperature program was set at 32 ºC for 5 min, ramped at 3 ºC min -1 to 220 ºC.The secondary oven temperature program was 10 ºC higher than the primary.The modulation period was 3 s with 0.75 s hot pulse duration and the modulator temperature was 30 ºC higher than the primary oven temperature.
GC conditions for solid samples: 50:1 split mode injection of 1 µL at 250 °C.Helium was used as carrier gas at a constant flow rate of 1.0 mL min -1 .The primary oven temperature program was: 70 ºC for 1 min, ramped at 3 ºC min -1 to 300 ºC.The secondary oven temperature program was 10 ºC higher than the primary.The modulation period was 3 s with 0.75 s hot pulse duration and the modulator temperature was 30 ºC higher than the primary oven temperature.
For all analyses the transfer line to the MS was at 280 ºC, the electron ionization at 70 eV, the mass range of m/z 35-600, the ion source temperature was 200 ºC, the detector was at -1700 V and the acquisition rate was 100 spectra s -1 .

Data processing for GC × GC-TOFMS
GC × GC-TOFMS data acquisition and processing were performed with ChromaTOF software version 4.0 (Leco, St. Joseph, MI, USA).Individual peaks were automatically detected on the basis of a 10:1 signal to noise ratio.Areas of the individual peaks were automatically acquired.

Results and Discussion
Comprehensive two-dimensional gas chromatography coupled to time of flight mass spectrometry (GC × GC-TOFMS) was employed to analyze unresolved material observed in 1 D-HTGC and evaluate the CO 2 effect on the distribution of n-alkanes, branched alkanes, alkenes and oxygen-containing compounds in the samples.
The selected column set had non-polar versus polar configuration, where separation in 1 D was based on the volatility while in 2 D the separation was based on the polarity of the compounds.4][25] However, some works have been published reporting GC × GC analyses for petrochemical samples with polar versus non-polar configuration. 26,27C × GC has been used for the analysis of FT products with configuration non-polar versus polar, 28 as well as polar versus non-polar configuration with GC × GC-TOFMS 29 and GC × GC-FID.30 1 D-HTGC analyses detected n-alkanes ranging from C 4 to C 15 for the liquid samples (FT1-L and FT2-L) and n-alkanes ranging from C 6 to more than C 50 for the solid ones (FT1-S and FT2-S).A typical 1 D-HTGC chromatogram of the FT-2S sample is presented in Figure 2, where unresolved compounds can be observed at the range C 9 to C 30 .
The detection of alcohols in FT samples is difficult because these compounds coelute in non-polar columns with branched alkanes, when analyzed by 1D-GC.This problem was solved by GC × GC-TOFMS.Figure 3 shows a twodimensional chromatogram for the FT1-L sample where two chromatographic bands can be seen, the lower band is formed by hydrocarbons and the upper one by n-alcohols which are more retained by the medium polar column (BPX-50) than hydrocarbons.The n-alcohols were the only oxygencontaining compounds detected in the samples.
The two-dimensional chromatograms were structured so that the compounds are resolved into groups.Each group  is formed by the n-alkanes, branched alkanes and alkenes with the same number of carbon atoms and the n-alcohol with less two carbon atoms, Figure 4a-c.The advantages of GC × GC can be observed even if concerning only the 1 D, as GC × GC provides thinner peak widths than conventional GC.In Figure 4b it can be observed that the separation of four branched alkanes had good resolution in 1 D.
In 1 D chromatography employing non-polar stationary phases with separation based on boiling point differences, the elution order is: the 1-alkenes elutes before the n-alkanes followed by the cis-2-alkene and the trans-2-alkenes. 30As the number of carbon increases, the cis-2-alkenes peak elutes closer to the n-alkanes peak, and in some cases they can be covered by the n-alkanes peak. 29This can lead to inaccurate determinations for n-alkanes and alkenes.This problem was not observed in GC × GC using polar versus non-polar configuration because alkanes and alkenes form two chromatographic bands in this case. 29,30uch problem was also not observed in this study either, Figure 4c shows the C 21 group where the resolution between the n-alkane and cis-2-alkene peaks can be seen; this resolution was effective throughout the whole chromatogram.Although alkanes and alkenes have similar retention times in the non-polar 1 D, the use of modulation provides higher resolution in GC × GC-TOFMS, even in 1 D.
All samples presented n-alkanes as a major class, Figure 5, whereas for solid samples the n-alkanes content represents 80-90% of the phase, for liquid samples n-alkanes it is responsible for 50-60% of the phase.
Unlike the solid samples, the liquid samples showed levels of alkenes higher than branched alkanes.The oxygencontaining compounds represent a minor class in both liquid (< 2%) and solid samples (< 0.5%).
The CO 2 effect on FT product composition was also evaluated.The samples obtained with CO 2, FT2-L (55%) and FT2-S (87%) presented lower levels of n-alkanes than the FT1-L (61%) and FT1-S (95%) samples.The process of branching of carbon chains was favored by the CO 2 , because the levels of branched alkanes were higher in samples FT2-L and FT2-S than in the samples obtained without CO 2 .
For both cases (FT synthesis with and without CO 2 ), the total amount of alkenes in the liquid phase, 28% (FT1-L) and 25% (FT2-L), was higher than in the solid phase, 1% (FT1-S) and 3% (FT2-S).Concerning the total amount of alcohols, were just observed few variations in both cases.
An observed effect of CO 2 in the FT product selectivity, for all classes, was that CO 2 favored the formation of compounds with lower molecular weights.In the sample FT2-L n-alkanes C 5 -C 9 represented 70% of all n-alkanes detected, while this range represented 50% of n-alkanes detected in FT1-L, Figure 6a.In the sample FT2-S n-alkanes C 9 -C 14 represented 44% of the overall composition of these compounds, while for FT1-S this range represented 27%, Figure 6b.
The effect of favoring the formation of compounds with lower molecular weights was also observed for branched alkanes and alkenes, but was more evident in the distribution of n-alcohols.In the liquid samples, the n-alcohols detected ranged from C 4 to C 12 , but n-pentanol and n-hexanol represented 60% of all n-alcohols detected in the sample FT2-L, while in the sample FT1-L these compounds represented 32% of alcohols detected.The range of n-alcohols detected for solid samples ranged from C 10 to C 19 and n-tridecanol was the major compound for FT1-S while for FT2-S the range was from C 7 to C 15  and n-nonanol was the major compound, which was not detected in FT1-L sample, Figure 6c-d.

Conclusions
GC × GC-TOFMS was able to resolve unresolved material as observed in the analysis of Fischer-Tropsch products by conventional GC.Higher levels of resolution for n-alkanes, branched alkanes, alkenes and n-alcohols analysis was provided by this technique.
The effect of 10% of CO 2 on the FT product selectivity was pronounced.The main class of products (n-alkanes, branched alkanes, alkenes and n-alcohols) shifted their carbon distribution to lower molecular weight compounds.
Although the products obtained in the presence of CO 2 have a lower average molecular weight when compared with FT synthesis without CO 2 , the n-alkanes are the major compounds in both products and the total amount of olefins and alcohols did not change significantly.Then, small additions of CO 2 on the Fischer-Tropsch process can be an environmentally attractive option, because it generates cleaner fuels with the CO 2 reuse.

Figure 2 .
Figure 2. One-dimensional chromatogram for FT2-S sample showing unresolved material which was resolved by GC × GC-TOFMS.

Figure 4 .
Figure 4. (a) Overview of the two-dimensional chromatogram for FT1-S sample: C 10 , n-decane; C 20 , n-eicosane; C 30 , n-triacontane.(b) Expansion of C n group showing the structuring of the compounds on the chromatographic plane, in this case n = 13.(c) Expansion of the C 21 group showing the resolution between cis-alkene isomer and n-alkane.

Figure 5 .
Figure 5.Comparison between the solid (a) and liquid (b) samples showing the relative proportion of n-alkanes, branched alkanes, alkenes and alcohols.