Diesel-length aldehydes and ketones via supercritical Fischer Tropsch Synthesis on an iron catalyst

Article abstract of DOI:10.1016/j.apcata.2010.07.032

Use of an iron-based catalyst (1 Fe:0.01 Cu:0.02 K by mole with and without 0.1 Zn) for Fischer Tropsch Synthesis with a supercritical hexanes reaction media resulted in a significant selectivity towards diesel-length aldehydes and methyl-ketones. Varying the residence time indicated that aldehydes are primary products that are converted by secondary reactions to olefins. Additionally, incorporation studies showed that octyl aldehyde and octyl alcohol incorporated into growing chains while the octyl olefin did not. The results of this work support n-alkanes and aldehydes as primary products, olefins as both primary and secondary products, CH₄ and CO₂ as secondary products, and an oxygenate mechanism for propagation.

Full text of DOI:10.1016/j.apcata.2010.07.032

Applied Catalysis A: General 386 (2010) 65–73
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Diesel-length aldehydes and ketones via supercritical Fischer Tropsch Synthesis
on an iron catalyst
Ed Durham, Sihe Zhang, Christopher Roberts∗
Department of Chemical Engineering, Auburn University, 210 Ross Hall, AL 36849, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 10 May 2010
Received in revised form 14 July 2010
Accepted 18 July 2010
Available online 22 July 2010
Use of an iron-based catalyst (1 Fe:0.01 Cu:0.02 K by mole with and without 0.1 Zn) for Fischer Tropsch
Synthesis with a supercritical hexanes reaction media resulted in a significant selectivity towards diesel-
length aldehydes and methyl-ketones. Varying the residence time indicated that aldehydes are primary
products that are converted by secondary reactions to olefins. Additionally, incorporation studies showed
that octyl aldehyde and octyl alcohol incorporated into growing chains while the octyl olefin did not. The
results of this work support n-alkanes and aldehydes as primary products, olefins as both primary and
secondary products, CH4 and CO2 as secondary products, and an oxygenate mechanism for propagation.
© 2010 Elsevier B.V. All rights reserved.
Keywords:
Fischer Tropsch Synthesis
Supercritical fluids
Iron catalyst
Aldehyde
Ketone
CO insertion
Mechanism
1. Introduction
LTFT (SC-FTS) to mitigate some of the problems seen in GP-FTS. A
number of benefits have been seen in SC-FTS relative to GP-FTS,
Fischer Tropsch Synthesis (FTS) is a process for converting syn-
gas (CO + H2) to fuels and chemicals. FTS can be performed at high
temperature (HTFT) on an iron catalyst for the production of gaso-
line and light olefins or at low temperature (LTFT) on either an iron
or cobalt catalyst for the production of diesel and wax [1]. Iron
is cheaper [1] and less active [2] than cobalt. Iron offers water–gas
shift activity, allowing for a lower usage ratio (the ratio of the rate of
consumption of H2 to the rate of consumption of CO) and a broader
range of the syngas ratio (the ratio of the feed rate of H2 to the
feed rate of CO) at the cost of a higher CO2 selectivity [3]. Iron also
provides greater resistance to poisoning by sulfur, but inferior resis-
tance to coking and attrition [3]. Iron gives a more olefinic product
and suppressed methane formation [4].
Sasol’s original LTFT reactor design utilized a fixed bed, as does
a shell design [5]. This form of operation is Gas Phase Fischer
Tropsch (GP-FTS). To overcome the limitations of GP-FTS (poor
extraction of products and heat from the catalyst, high pressure
drop, and poor economies of scale), Sasol developed their slurry
bed reactor (SP-FTS) [5]. Kaoru Fujimoto’s group [6] pioneered the
use of a supercritical fluid as the reaction medium for fixed bed
including suppressed methane formation [6–12], CO2 formation
[8,12] and improved activity maintenance [10]. Additionally, many
researchers have seen an enhancement in the olefin selectivity into
higher carbon numbers [8,9,12–15]. This can be explained by the
supercritical media having a high capacity to extract and stabilize
products that would otherwise undergo secondary reactions (such
as olefin hydrogenation to paraffins).
The mechanism of the Fischer Tropsch reaction is still a mat-
ter of contention, with Claeys and van Steen listing four pathways
[16]: the alkyl mechanism, the alkenyl mechanism (associated with
Maitlis), the enol mechanism (associated with Storch), and the CO
insertion mechanism (via coordination chemistry and homogenous
catalysis). Three key aspects of any Fischer Tropsch Mechanism are:
(1) the pathway for CO dissociation (whether it is unassisted – prior
to hydrogenation. or assisted – after some hydrogenation), (2) the
structure of the initiator species, and (3) the structure of the propa-
gator species. See Table 1 for a presentation of the four mechanisms
listed above.
For non-oxygenate mechanisms, the formation of oxygenates
is often explained as being due to a different termination step,
usually CO insertion [4,16,17]. In preliminary work, we observed
a significant selectivity to diesel-length aldehydes on an iron–zinc
catalyst promoted with copper and potassium. The remainder of
this study was undertaken to determine the role of these products
in the reaction pathway.
∗
Corresponding author. Tel.: +1 334 844 2036; fax: +1 334 844 2063.
E-mail address: croberts@eng.auburn.edu (C. Roberts).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.07.032

66
E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
Table 1
Role of hydrogen in CO dissociation, initiator, and propagator for four proposed FTS mechanisms. Information from Ref. [16].
Mechanism
Alkyl
Alkenyl
Enol
CO insertion
CO Dissociation
Initiator
Propagator
Unassisted
RCH2
CH2
Unassisted
RCHCH
CH2
Assisted
RCOH or RCHOH
CHOH
Assisted
RCH2
CO
2. Experimental
For every experiment, the calcined catalyst was activated
(reduced) in situ using syngas (50 SCCM/g) at 270 ◦C.
2.1. Catalyst synthesis
2.2. Apparatus
Three batches of catalyst were made and used in this study (all
compositions are molar):
Two reactor systems were used in this study: a fixed bed reactor
for GP-FTS and SC-FTS and a stirred Parr Reactor for SP-FTS.
A schematic of the fixed bed reactor system is shown in Fig. 1.
In this system, syngas (A) and helium (B, when appropriate) are fed
through a mass flow controller (C) and mixed with hexanes (D, for
SC-FTS) fed through a positive displacement HPLC pump (E). The
mixture passes through heated lines to the reactor (F, downflow)
in a split tube furnace (G). From the reactor, the effluent passes
through heated tubing to the hot trap (H, maintained at 200 ◦C to
condense heavy wax, ideally without significant partitioning below
C20). The hot trap condensate is allowed to accumulate and can be
periodically collected and analyzed. The vapor from the hot trap
passes through heated tubing to the BPR (I, back pressure regu-
lator, used to maintain the reactor pressure), and through heated
tubing to the cold trap (J, maintained at slightly sub-ambient tem-
perature). Liquids are allowed to accumulate in the cold trap and
are periodically collected and analyzed via manual injection to the
FID-GC. The gasses from the cold trap pass to a six-way injection
valve (K) used to automatically inject them to the TCD-GC. From
there the gasses are vented to the fume hood.
The reactor in this system is an HIP microreactor with a length
of 10 inches and an ID of 0.5 in. The upper half has been reamed
out to an ID of 0.625 in. A stainless steel filter disk rests on the rim
between the larger and smaller ID sections, with the catalyst resting
upon the disk. The thermocouple is a 6-point profile thermocouple
(Omega Engineering PP6-36-K-G-18) with the third junction in the
catalyst bed and used to control the furnace.
A schematic of the Parr Reactor is shown in Fig. 2. In this sys-
tem, syngas (A) and helium (B, when appropriate) are fed through a
mass flow controller (C), then through a heated line into the reactor
(D) where they are bubbled through the agitated (600 RPM) slurry.
Vapor is pulled off of the top of the reactor and passes through a
heated line to the cold trap (E, maintained at sub-ambient temper-
ature). The condensate in the cold trap is allowed to accumulate to
be periodically collected and analyzed by manual injection to the
FID-GC. The vapor from the cold trap passes through a BPR (F, used
to maintain the reactor at the desired pressure), then the 6-way
valve (G, used to automatically inject to the TCD-GC), and finally is
vented to the fume hood. Positioning the BPR after the cold trap cre-
ates some analytical problems (loss of light products during liquid
collection), but improves the reliability of the system. The reactor
was pre-filled with 500 mL of wax which was slurried with 4 g of
iron catalyst.
A 1.0 Fe:0.10 Zn:0.02 K:0.01 Cu
B 1.0 Fe:0.10 Zn:0.02 K:0.01 Cu
C 1.0 Fe:0.01 Cu:0.02 K
The batches were synthesized by a procedure similar to one
published by Enrique Iglesia’s group [18]. The materials for this
procedure are as follows (all listed here and elsewhere used as
received):
Water: DIUF Water (Fisher W2-4).
Ethanol: Absolute Ethyl Alcohol (Pharmco-Aaper E200).
INH: ACS Reagent Grade Iron (III) Nitrate Nonahydrate
(Sigma–Aldrich 216828-500G).
ZNH: Reagent Grade Zinc Nitrate Hexahydrate (Sigma–Aldrich
228737-100G).
AC: ACS Reagent Grade Ammonium Carbonate (Sigma–Aldrich
207861-500G).
CNH: ACS Reagent Grade Copper (II) Nitrate Hydrate
(Sigma–Aldrich 223395-100G).
PC: ACS Reagent Grade Potassium Carbonate (Sigma–Aldrich
209619-100G).
For each batch, a stock iron solution (1 M INH and 0.1 M ZNH
in water for Batch A and Batch B, 1 M INH for Batch C) and a stock
reducing solution (AC in water – saturated at room temperature)
were made.
At a controlled rate (2 mL/min for Batches A and Batch C,
6 mL/min for Batch B), the iron solution was added to water (50 mL
for Batch A and Batch C, 150 mL for Batch B) maintained at 80 ◦C
using an Eldex Laboratories Syringe Pump (B-100-S). The reduc-
ing solution was added manually to maintain the pH at 7.0 (as
measured by a Denver Instrument UB-10 pH meter) with vigorous
stirring (via magnetic stir bar). At the end of the co-precipitation
(15 min for Batch A, 45 min for Batch B, 90 min for Batch C), reagent
addition was stopped and the solution allowed to cool with con-
tinued stirring. The slurry was then vacuum filtered. The filter cake
was re-slurried in water and then re-filtered 4 times. The filter cake
was then re-slurried in ethanol and then vacuum filtered twice. The
filter cake was then dried overnight at 80 ◦C.
The dried catalyst was then calcined at atmospheric pressure
in flowing air, the temperature ramped at 5 ◦C/min and held at
400 ◦C for 240 min. When the catalyst had cooled, the pore volume
was determined by the addition of water (measured by mass). The
catalyst was then dried again at 80 ◦C.
The potassium (PC) and copper (CNH) promotion were done by
incipient wetness impregnation, with the catalyst dried overnight
at 80 ◦C and calcined in flowing air with the temperature ramped
at 5 ◦C/min to 400 ◦C and held for 4 h after each impregnation. For
Batch A and Batch B, the potassium impregnation was done prior
to copper impregnation. For Batch C, the copper impregnation was
done first.
The syngas was purchased pre-mixed from Airgas (Certified
Standard) with 1.5% N2 (internal standard) and H2/CO = 1.65. The
helium is UHP grade and is used for pressure testing, startup, and
shut-down. The hexanes (the media used during SC-FTS) was pur-
chased in bulk from Fisher Scientific. The paraffin wax (the media
used during SP-FTS) was purchased from Fisher Scientific (CAS
8042-47-5). It is a complex mixture of unknown components that
elutes from the FID-GC between the C22 and C35 n-paraffins.
Analysis of the cold trap vapor stream was done by online injec-
tion (Valco 6-way valve) to a Varian 3800 GC with a TCD detector

E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
67
Fig. 1. Schematic of the fixed bed reactor system (used for GP-FTS and SC-FTS). (A) Syngas cylinder, (B) helium cylinder, (C) mass flow controller, (D) hexanes bottle, (E) HPLC
pump, (F) reactor, (G) furnace, (H) hot trap, (I) BPR, (J) cold trap, (K) 6-way valve.
and a Hayesep column (Grace Davison 2836PC). Standards were
used to determine the response factors of the various components
and the N2 internal standard was used to determine the rate of the
cold trap vapor stream. Analysis of the cold trap liquids and Parr
reactor residual liquids was done by manual injection to a Varian
3300 GC with an FID detector and a DB-5 column (Agilent 125-
5032). The mass response factors were assumed to be identical for
all pure hydrocarbon components (olefins, paraffins, and isomers)
and determined by standard injection for other classes of products
(alcohols and aldehydes).
The GC/MS analysis was performed at the Auburn University
Mass Spec Center. The analysis utilized a Waters 6890N GC with
a DB5-MS column (Cat# 1225532, J&W Scientific) coupled to a
Time of Flight Mass Analyzer (GCT Premier, Waters). Component
identification was done by comparing the electron impact fragmen-
tation pattern (70 eV) with those in the compound library (NIST
2003). Compounds with Match and Reverse Match scores above
800 and probabilities above 90% were selected as matches. The
identification was confirmed by elemental composition analysis
using accurate mass measurement with an internal calibrant (lock-
mass 218.9856 m/z, heptacosafluorotributylamine, Sigma) with an
acceptable error of less than 5 ppm and by isotope modeling com-
paring the experimental and theoretical isotope distribution. The
component identifications from the GC/MS were also confirmed by
retention time of standard compounds on the FID-GC. Quantitative
analysis was done on the FID-GC with response factors determined
though the analysis of corresponding standards.
2.3. Procedure
First study (GP-FTS and SC-FTS study): the first study used 2 g of
the Batch A catalyst in the fixed bed reactor system. After a brief (ca.
10 h) period under GP-FTS conditions the system was converted to
SC-FTS operation. After a series of system upsets due to mechanical
issues, the reactor was stabilized and a brief study was conducted
at various syngas rates (150 SCCM syngas, 100 SCCM, 50 SCCM,
300 SCCM, 150 SCCM, sequentially), with the media (hexanes) to
syngas ratio kept at 1 mL/min per 50 SCCM (approximately 3.5 mol
media per mol syngas). During GP-FTS operation, the pressure was
maintained at 17.5 bar. During SC-FTS operation, the pressure was
maintained at 75 bar (maintaining the syngas partial pressure at
17.5 bar). With the exception of the catalyst activation (done at
Fig. 2. Schematic of the Parr Reactor (used in the SP- FTS). (A) Syngas cylinder, (B) helium cylinder, (C) mass flow controller, (D) reactor, (E) cold trap, (F) BPR, (G) 6-way
valve.

68
E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
Fig. 4. Second study (GP-FTS)—ASF plot (2 g catalyst, 100 SCCM syngas, P = 17.5 bar,
T = 240 ◦C, XCO = 45%).
Fig. 3. First study (SC-FTS)—ASF Plot (2 g catalyst, 150 SCCM syngas, 3 mL/min hex-
anes, P = 75 bar, T = 240 ◦C, XCO = 70%).
270 ◦C and 1 bar), the reaction temperature was maintained at
240 ◦C.
Second study (GP-FTS and SP-FTS comparison): the second study
used the Batch B catalyst, testing it both for GP-FTS and SP-FTS.
In the GP-FTS study 2 g of catalyst were loaded into the fixed bed
reactor. Once conversion had stabilized, the syngas rate was varied
(100 SCCM, 200 SCCM, 100 SCCM, 50 SCCM, sequentially) to deter-
mine the reactor performance over a broad range of conversion
levels over the 150 h time on stream. The experiment was done
at 240 ◦C. For the SP-FTS portion of the second study, the reac-
tor was loaded with 500 mL of paraffin wax and 4 g of catalyst.
After the reduction, the reactor was briefly run at double rate (syn-
gas rate = 400 SCCM) and 270 ◦C, then regular rate (200 SCCM) and
270 ◦C before going to the regular rate (200 SCCM) and temperature
(240 ◦C). For both the GP-FTS and SP-FTS the pressure was held at
17.5 bar.
Third study (SC-FTS): the third study used the Batch C catalyst,
diluting 1 g of catalyst in 10 g of acid washed and calcined silica sand
(Acros Organics 370940010) under SC-FTS conditions. After allow-
ing the system performance to stabilize, the syngas rate was varied:
100 SCCM, 50 SCCM, 100 SCCM, 200 SCCM, then 100 SCCM to test
selectivity at moderate, high, and low conversion levels (the hex-
anes rate was maintained at 1 mL/min/50 SCCM). After the study of
the selectivity at different conversion levels, the incorporation of
aldehydes, alcohols, and olefins were studied.
Fig. 5. Second study (SP-FTS)—ASF plots (not normalized) (4 g catalyst, 200 SCCM,
P = 17.5 bar, T = 240 ◦C, XCO = 65% for cold trap vapor and cold trap liquid. Products
from whole run for reactor residual).
With the possible exception of GP-FTS, each of the ASF plots
shows a 2-alpha distribution with the two regions meeting in the
low teens. For this work, the most important observation is that the
high-carbon number alpha values are consistent among the differ-
ent catalyst batches and reaction media. Additionally, the ASF plots
for different conversion levels under GP-FTS and SC-FTS are not
shown, but there is little influence of conversion level on the ASF
plots.
3.2. Gas product analysis
In the incorporation portion of this study, the system was first
run under normal SC-FTS conditions to give a baseline. The media
was then doped with octyl aldehyde (Acros Organics 129481000).
After returning to baseline, fresh media was doped with octyl alco-
hol (Aldrich 47232-8). Following another return to baseline, the
fresh media was doped with 1-octene (Acros Organics 301250025),
and the study concluded with a return to baseline. Each portion
of the incorporation half of this study lasted approximately 12 h
with each dopant fed at a concentration of 0.20 mol per mol CO.
The effects of the attempted incorporation were determined by
comparing the conversion and amounts of various products seen
during attempted incorporation with that seen during the baseline
preceding and following it.
The crossplot of CO2, CH4, and C3 olefin selectivity versus CO
conversion for the first study is presented in Fig. 7. The CO2 selec-
tivity is calculated on a CO basis [CO2 generated/CO consumed],
the methane selectivity on a CO2-free basis [CH4 generated/(CO
consumed − CO2 generated)], and the C3 olefin selectivity on a
C3 hydrocarbon basis [propene/(propene + propane)]. Operation
3. Results
3.1. Overall product distribution
The overall product distribution for the first study (SC-FTS –
Fig. 3), the second study (GP-FTS – Fig. 4 and SP-FTS – Fig. 5), and
the third study (SC-FTS – Fig. 6) are as follows. The slurry-phase
ASF plot has the three separate product fractions (cold trap gasses,
cold trap liquids, and reactor residual liquids).
Fig. 6. Third study (SC-FTS)—ASF plot (1 g catalyst, 200 SCCM syngas, 4 mL/min hex-
anes, P = 75 bar, T = 240 ◦C, XCO = 45%).

E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
69
Fig. 7. First study (SC-FTS (empty, green in web) and GP-FTS (shaded, red in
web))—CO2, C3 Olefin, and CH4 vs. CO Conversion.
Fig. 10. First study (SC-FTS)—breakdown of liquid product by type –v– carbon
number (2 g catalyst, 150 SCCM syngas, 3 mL/min hexanes, P = 75 bar, T = 240 ◦C,
XCO = 70%, propagation probability = 87%).
Fig. 8. Second study (GP-FTS (empty, red in web) and SP-FTS (full, blue in
web))—CO2, C3 Olefin, and CH4 vs. CO Conversion.
Fig. 11. Second study (GP-FTS)—breakdown of liquid product by type –v– carbon
number (2 g catalyst, 100 SCCM syngas, P = 17.5 bar, T = 240 ◦C, XCO = 45%, propaga-
tion probability = 86%).
under SC-FTS conditions suppressed the methane and CO2 selec-
tivity significantly and the C3 olefin selectivity slightly relative to
GP-FTS. The methane and CO2 selectivity both decreased markedly
with decreasing CO conversion, appearing to extrapolate to zero
as the CO conversion approaches zero and the product distribu-
tion becomes the primary product distribution. Additionally, the C3
olefin selectivity appeared to drop as the CO conversion decreased,
though not appearing to extrapolate to zero as CO conversion
approached zero, indicating that propene is likely both a primary
and secondary product.
and a large decrease in CH4 selectivity. The decrease in methane
selectivity in going from GP-FTS to SP-FTS is comparable to that
seen in going from GP-FTS to SC-FTS.
The crossplot of CO2, CH4, and C3 olefin selectivity versus CO
conversion for the third study is presented in Fig. 9. The SC-FTS
results seen in this study (with a zinc-free catalyst) match those
seen in SC-FTS operation in the first study. Methane and CO2 again
appear to be secondary products while propene appears to be both
a primary and secondary product.
The crossplot of CO2, CH4, and C3 olefin selectivity versus CO
conversion for the second study is presented in Fig. 8. Under GP-FTS
operation, there was no trend in CO2, CH4, or C3 olefin selectivity
versus CO conversion. Going from GP-FTS to SP-FTS gave a mild sup-
pression in C3 olefin selectivity, a mild increase in CO2 selectivity,
3.3. Liquid product breakdown by type
Thebreakdown ofthe liquidproductbytypeversus carbonnum-
ber for the first study during SC-FTS operation is shown in Fig. 10
(the same liquid sample used to generate the ASF plot shown in
Fig. 3). With the exception of the “Other” products gradually disap-
pearing, there is little influence of carbon number on the relative
amounts of olefins, paraffins, and aldehydes (the aldehyde identi-
fication was confirmed by GC retention time and GC–MS analysis).
The analytical procedure used in this study lumped the alcohols
into the “Other” category. Branching has been shown to decrease
with increasing carbon number [4].
Thebreakdown ofthe liquidproductbytypeversus carbonnum-
ber for the second study is shown in Fig. 11 (GP-FTS operation – the
same sample used for the ASF plot in Fig. 4) and Fig. 12 (SP-FTS
operation cold trap liquids – the same liquid sample used for the
ASF plot in Fig. 5).
The GP-FTS breakdown showed different behavior than that
seen in Fig. 10 with SC-FTS. First, the liquid aldehydes were not
present in GP-FTS in detectable quantities while internal olefins
were observed. Second, with increasing carbon number the GP-FTS
Fig. 9. Third study (SC-FTS)—CO2, C3 Olefin, and CH4 vs. CO Conversion.

70
E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
Fig. 13. Third study (SC-FTS)—breakdown of liquid product by type –v– carbon
number (1 g catalyst, 200 SCCM syngas, 4 mL/min hexanes, P = 75 bar, T = 240 ◦C,
XCO = 45%, propagation probability = 86%).
Fig. 12. Second study (SP-FTS)—breakdown of liquid product by type –v– carbon
number (4 g catalyst, 200 SCCM syngas, P = 17.5 bar, T = 240 ◦C, XCO = 65%, propaga-
tion probability = 84%).
Fig. 14. First study (SC-FTS)—breakdown of liquid product by type –v– conversion
level (P = 75 bar, T = 240 ◦C, CO conversion in parenthesis).
liquid product became much less olefinic and more paraffinic, indi-
cating the hydrogenation of olefins to their corresponding paraffins.
The SP-FTS product breakdown by type, like that in the first
study, did not isolate alcohols from the “Other” category. As with
GP-FTS, no liquid aldehydes were seen in this SP-FTS study while
significant quantities of internal olefins were detected. Like SC-FTS,
the product distributions in the cold trap liquids and reactor resid-
ual liquids were independent of carbon number, apart from the
gradual disappearance of the “Other” products. The cold trap liquids
in SP-FTS were far more olefinic than the reactor residual liquids,
which is consistent with the reactor residual liquids having a far
higher contact time. A greater fraction of the reactor residual liq-
uids are present as “Other” products than in the cold trap liquids.
Assuming that these products are predominately branched prod-
ucts, the higher selectivity in the reactor residual liquids supports
branched compounds as secondary products, likely formed from
olefins. The apparent disappearance of internal olefins at the end of
the diesel range is due to analytical limitations at the corresponding
high retention times for these compounds.
The breakdown of liquid product by type versus carbon num-
ber for SC-FTS on a zinc-free iron catalyst from the third study is
shown in Fig. 13. This SC-FTS study gave very similar results to those
obtained from the first SC-FTS study (Fig. 10), with little change
versus carbon number apart from the gradual disappearance of
“Other” products. It is important to note that the analytical proce-
dures employed during the third study were improved to allow for
the detection of alcohols and methyl-ketones (the methyl-ketone
identification was confirmed by GC retention time and GC–MS anal-
ysis). As in the first study, significant quantities of liquid aldehydes
were detected in this SC-FTS investigation. A significant quantity of
methyl-ketones was also found.
is similar to that of the paraffin of two greater carbon number. At
a low carbon number (C8), the methyl-ketone elutes prior to the
olefin. At higher carbon numbers (C9–C12) the methyl-ketone is
indistinguishable from the olefin. The methyl-ketone can again be
observed from C13 to C14 between the paraffin and the olefin. The
methyl-ketone peak then merges with the paraffin peak from C15
to C18. Above C19 the methyl-ketone elutes after the paraffin and
is observed until it disappears into signal noise. It is prohibitively
likely that the methyl-ketones are present at the carbon numbers
where the peak cannot be resolved with selectivities similar to the
carbon numbers where they are resolvable.
The breakdown of the liquid products by type versus conversion
level for the first study under SC-FTS conditions is shown in Fig. 14
with the analysis limited to the C11–C17 range. As CO conversion
decreased and the product spectrum approached the primary prod-
uct spectrum, the aldehyde selectivity increased markedly while
the “Other” and olefin selectivity decreased. The paraffin selectiv-
ity was largely unaffected by conversion level. This indicates that
aldehydes are primary products. The “Other” products appear to
be secondary products. The olefins appear to be secondary prod-
ucts as well, but it is doubtful that the olefin selectivity would go
to zero if the CO conversion was extrapolated to zero, indicating
that olefins are both primary and secondary products. Paraffins
appear to be primary products, and the hydrogenation of olefins to
paraffins appears to be effectively suppressed under supercritical
conditions, as evidenced by the lack of an increase in paraffinicity
with increasing carbon number (Fig. 10) or increasing CO conver-
sion (Fig. 14).
The breakdown of the liquid products by type versus conver-
sion level for the second study under GP-FTS conditions is shown in
Fig. 15 for the C10–C17 range. With decreasing CO conversion, here
the product becomes less paraffinic and more olefinic. This sup-
In Fig. 13, it appears that methyl-ketones are only present at
C8, from C13 to C14, and >C18. However, this sporadic appearance
is due to analytical limitations. The methyl-ketones retention time

E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
71
Fig. 17. Third study (SC-FTS)—breakdown of liquid product by type –v– conversion
level (P = 75 bar, T = 240 ◦C, CO conversion in parenthesis).
Fig. 15. Second study (GP-FTS)—breakdown of liquid product by type –v– conver-
sion level (P = 17.5 bar, T = 240 ◦C, CO conversion in parenthesis).
selectivity to liquid aldehydes, minimal influence of carbon number
on the product type breakdown, an increase in aldehyde selectivity
with decreasing CO conversion, a decrease in olefin selectivity with
decreasing CO conversion, and no influence of CO conversion on the
paraffin selectivity) were obtained in another SC-FTS study using a
1 Fe:0.1 Zn:0.01 Cu:0.02 K by mol catalyst and another study using
Batch C without a hot trap in the reactor system.
3.4. Aldehyde, alcohol, and olefin incorporation
After the completion of the third study presented above, an addi-
tional investigation was undertaken to explore the incorporation of
added C8 aldehyde, C8 alcohol, and 1-octene in the Fischer Tropsch
reaction. The incorporant was added to the hexanes media such that
it was present at the reactor inlet at a rate 20% of that of the CO in the
syngas. The analysis was performed by comparing the performance
during the attempted incorporation (conversion of CO and H2 and
the production rate of the different products) with that observed
during a period of SC-FTS operation without an incorporant imme-
diately preceding and following the attempted incorporation. C8
aldehyde incorporation was studied first, followed by C8 alcohol
incorporation, and finally 1-octene incorporation.
The addition of the C8 aldehyde resulted in some incorporation
into heavier products with a selectivity comparable to that seen in
the baselines (i.e. no incorporant added). There was no distinguish-
able effect of the C8 aldehyde on conversion, methane selectivity,
or CO2 selectivity. The formation of C2 and C3 products appear to
have been suppressed. In addition to incorporation, the C8 alde-
hyde was converted to the C7 paraffin, C8 olefin, C8 alcohol, C9
methyl-ketone, C9 alcohol, and an unknown C15 oxygenate.
The addition of the C8 alcohol resulted in some incorporation
into heavier products with a selectivity comparable to that seen in
the baselines. There was no distinguishable effect of the C8 alcohol
on conversion, methane selectivity, CO2 selectivity, C2 selectivity,
or C3 selectivity. In addition to the incorporation, the C8 alcohol
appears to have been converted to the C8 aldehyde, C9 alcohol, and
C9 aldehyde.
Fig. 16. Third study (SC-FTS)—breakdown of liquid product by type –v– conversion
level (P = 75 bar, T = 240 ◦C, CO conversion in parenthesis).
ports olefins as primary products that are converted by secondary
reactions to paraffins, contrary to what was seen in SC-FTS opera-
tion. We believe this to be due to the failure the GP-FTS media to
inhibit the consumption of the aldehyde or aldehyde-like reaction
intermediate. Consequently, because the consumption of aldehy-
des is nearly complete even at low conversion levels in GP-FTS, with
increasing CO conversion the secondary reaction that is observed is
the hydrogenation of the olefins giving a false view of the primary
product spectrum.
The breakdown of the liquid products by type versus conver-
sion level for the third study under SC-FTS conditions is shown in
Fig. 16 for the C11–C17 range. This data confirms much of what was
seen in the first study in SC-FTS (Fig. 14) while adding additional
information, with aldehydes appearing to be primary products that
are converted via secondary reactions to alcohols and olefins. It
appears that olefins would have a significant selectivity even with
a CO conversion approaching 0%, making it likely that they are both
primary and secondary products. Paraffins appear to be primary
products and their formation through secondary reactions appears
to be suppressed.
The breakdown of the liquid products by type versus conver-
sion level for the third study under SC-FTS conditions is shown
in Fig. 17 for the C20–C24 range. While “Other” products are no
longer distinguishable from the chromatogram noise and the alco-
hol peak can no longer be distinguished from the olefin peak, the
olefin, paraffin, methyl-ketone, and aldehyde peaks are still distin-
guishable. This data again confirms aldehydes as primary products
that are partially converted to olefins, olefins as both primary and
secondary products, methyl-ketones as likely secondary products,
and paraffins as primary products.
The addition of 1-octene resulted in no discernible incorpora-
tion into heavier products. There was no change in CO conversion,
though a drop in usage ratio was observed. The methane, CO2, C2
and C3 selectivities were unchanged. The 1-octene was converted
to C8 internal olefins, the C8 paraffin, and a product that is likely
the C10 olefin.
4. Discussion
The observations reported here for SC-FTS have been replicated
in our laboratory multiple times, but, for simplicity, only selected
experiments are reported here. The same results (i.e., significant
Low temperature Fischer Tropsch produces two classes of prod-
ucts: hydrocarbons and oxygenates. The predominant hydrocarbon
products are n-paraffins, n-olefins (mostly terminal), and branched

72
E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
Table 2
paraffins and olefins. Minor hydrocarbon products include aro-
matics and dienes. The predominant oxygenates are aldehydes
(mostly linear), alcohols (mostly terminal and linear), ketones
(with the carbonyl group mainly on the ␤-carbon), carboxylic acids
(mostly linear), and esters (mostly linear). Minor oxygenate prod-
ucts include acetals, ethers, furans, and phenols [19].
Simulation of the thermodynamics of aldehyde, alcohol, and olefin inter-conversion
at SC-FTS conditions.
Feed (mol/h)
Product (mol/h)
Hexane
Hydrogen
Water
Octyl aldehyde
1-Octyl alcohol
1-Octene
3.5
3.5
0.24
0.27
0.25
0.25
0.01
0.01
0.01
Paraffins are shown to be primary products by varying the res-
idence time (conversion level) under SC-FTS conditions. There are
several possible pathways for their formation. Burt Davis’ group
observed theconversionofadded alcohols to theonecarbonshorter
paraffin in a process that also formed CO2 [20] in a slurry-phase
reactor. However, under SC-FTS conditions, we did not observe the
formation of the one shorter paraffin from an added alcohol, but did
observe this phenomenon with the added aldehyde. The conver-
sion of the aldehydic intermediate to a 1-carbon shorter paraffin is
a possible pathway to the primary formation of paraffins. However,
the apparent absence of CO2 as a primary product, the absence of
methane as a primary product, and the absence of heavier paraffins
as secondary products make us consider this unlikely. As such, we
prefer the alpha-hydrogenation of a surface alkyl group as a more
plausible mechanism for the primary production of paraffins.
The selectivity to diesel-length olefins decreased with decreas-
ing conversion under SC-FTS conditions, contrary to what was seen
on the same catalyst under GP-FTS conditions and to what we have
observed previously in SC-FTS on a cobalt catalyst. The conversion
of octyl aldehyde to 1-octene was observed in the incorporation
study, confirming the conversion of aldehydes to olefins as evi-
denced by the changes is selectivity with variations in conversion.
However, the olefin selectivity does not appear to approach zero
as CO conversion extrapolates to zero, suggesting that olefins are
also primary products. Beta hydrogen elimination from a surface
alkyl group is a plausible mechanism for the primary production of
olefins. No discernable internal olefins were detected under SC-
FTS operation, though internal olefins were seen in GP-FTS and
SP-FTS, supporting internal olefins as only secondary products.
Additionally, 1-octene, when added to SC-FTS, did form some inter-
nal octenes, supporting this conclusion.
4.5 × 10−7
6.1 × 10−5
0.030
reactor at typical SC-FTS conditions (P = 75 bar, T = 240 ◦C) where
the aldehyde, alcohol, and olefin were free to be interconverted to
minimize the Gibbs Free Energy of the system. The results of this
simulation are shown in Table 2. The simulation indicates that the
conversion of aldehydes to alcohols and alcohols to olefins should
be, under typical SC-FTS conditions, near total. Consequently, given
the high ratio of aldehydes to alcohols observed, we conclude that
aldehydes are primary over alcohols. The fact that added 1-octene
did not react to give an aldehyde further establishes that the alde-
hydes are not secondary products of olefins, but that the reverse
is true. Additionally, in 1962, Wender [22] demonstrated that the
synthesis of an aldehyde from an olefin of one fewer carbon number
with cobalt hydrocarbonyls readily occurs.
Alcohols are likely secondary products from the hydrogenation
of aldehydes given the two products’ capacity for inter-conversion,
the drop in alcohol selectivity with decreasing CO conversion, and
their capacity to be converted to heavier products.
Methyl-ketones appear to be secondary products formed from
aldehydes, though they have been asserted to be formed via car-
boxylic acid decomposition [19]. The pseudo-dimerization product
of the added C8 aldehyde to a C15 oxygenate may be a ketone, but
it is not a methyl-ketone [23].
Our analytical system does not allow for the detection of car-
boxylic acids or esters. The inability to measure carboxylic acids
is particularly unfortunate since formate-type species have been
proposed as being the chain initiator on iron [20].
Branched compounds have been shown to be formed from the
incorporation of non-terminal alcohols [20]. The high selectivity to
‘Other’ products (presumed to be predominantly branched com-
pounds) in the reactor residual liquids in SP-FTS also suggests their
formation from olefins. While the incorporation of 1-octene was
not seen in this work, it has been generally observed, albeit at
a lower rate than that for alcohols [20]. A terminal olefin should
allow for initiation at the 1 carbon (giving an n-paraffin) or 2 car-
bon (giving a methyl-branched product). A 2-olefin should allow
for initiation at the 2 carbon (giving a methyl-branched product)
or the 3 carbon (giving an ethyl-branched compound). The forma-
tion of methyl-ketones from aldehydes offers a pathway beside
isomerization to get from a terminal species to a methyl-branched
one. However, under supercritical conditions methyl-ketones are
present at carbon numbers into the wax product range while
branched products die out in the diesel range.
Aldehydes are demonstrated to be primary products by varying
the residence time in SC-FTS. The incorporation study demon-
strated their capacity to incorporate into heavier products. Burt
Davis’ group has shown that incorporated alcohols act as chain
initiators [20]; we believe that this is likely to also be the means
of incorporation for aldehydes. Aldehydes have elsewhere been
shown to give incorporation behavior comparable to that of alco-
hols [21]. Aldehydes and alcohols appeared to interconvert in the
incorporation study, suggesting that either could be the primary
and either the secondary product. To study this uncertainty, a sim-
ulation was done in AspenPlus using the Peng Robinson equation of
state. In this simulation, a stream of hexane, hydrogen, water, octyl
aldehyde, 1-octyl alcohol, and 1-octene were fed to an equilibrium
The high selectivity to aldehydes suggests an oxygenate mech-
anism (likely CO Insertion [4,16,17]) as, if nothing more,
a
termination mechanism for the reaction. However, the capacity for
an aldehyde to be converted to possibly all of the reaction product
types suggests a broader role for CO insertion. The fact that the alde-
hyde can also initiate chain growth suggests the oxygenate process
as the growth mechanism for the reaction as well. The broad spec-
trum of apparent primary products suggests that there are multiple
types of surface species. However, the consistency of the product
type with carbon number suggests sequential termination path-
ways as opposed to parallel mechanisms. As such, we believe the
initiator to be fundamentally different from the intermediate (the
product of the initiator reacting with the propagator). The specific
identity of these two species is still a matter of inquiry. We believe
a CO Insertion mechanism to be more likely given the product spec-
trum than an enol mechanism. Several variations on CO insertion
have been proposed [24].
The mechanism for CO dissociation in Fischer Tropsch Synthe-
sis is still a matter of dispute. However, we believe that the case
for hydrogen assisted dissociation has been compellingly made
[25–27]. Additionally, if hydrogen assisted dissociation is accepted,
an oxygenate mechanism in general and CO insertion specifically
becomes highly plausible (see Fig. 18).
A C14 labeling incorporation study would be very useful to bring
clarity here. Enrique Iglesia’s group [17] has made a very strong
case that the behavior of an added incorporant does not necessar-
ily match that of the same component produced in situ. Our group
has carried out a number of investigations [9,10,12,28] of SC-FTS on
a cobalt catalyst and we have not observed diesel-length aldehydes.

E. Durham et al. / Applied Catalysis A: General 386 (2010) 65–73
73
Fossil Fuel Science, and the Sun Grant Consortium. We would like to
express gratitude to Dr. Yonnie Wu, Director of the Auburn Univer-
sity Mass Spec Center, for invaluable assistance in performing the
GC/MS experiments and the interpretation of their results. Addi-
tionally, we would like to thank Brian Schweiker and Mike Hornsby
for their help with instrumentation and equipment, Mario Eden
(Auburn University) for his assistance with the Aspen simulation,
Mohindar Seehra (West Virginia University) for catalyst characteri-
zation, and Enrique Iglesia (University of California at Berkeley) and
Andre Steynberg (Sasol) for technical correspondence on specific
questions.
Fig. 18. Presentation of (a) the hydrogenation of CO prior to dissociation (hydrogen
assisted dissociation) and (b) CO Insertion. No claim as to the nature of the bond
between the CO and the catalyst is intended.
Appendix A. Supplementary data
There are two possibilities here. The first is that cobalt-based FTS
has a different mechanism that would not produce aldehydes as pri-
mary products. The second is that, even if cobalt-based FTS utilizes
the same mechanism as iron, cobalt is a much stronger hydrogena-
tion catalyst than iron [29], so the aldehydes would be too rapidly
converted to olefins and paraffins to be observed at higher carbon
numbers. As such, while we feel based on the data presented here
that CO insertion in some form is the chain growth mechanism for
FTS on an iron catalyst, we cannot make any claims for cobalt.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.apcata.2010.07.032.
References
[1] M.E. Dry, Catal. Today 71 (2002) 227–241.
[2] F. Morales, B.M. Weckhuysen, Catalysis 19 (2006) 1–40.
[3] A.Y. Khodakov, W. Chu, P. Fongarland, Chem. Rev. 107 (2007) 1692–1744.
[4] M.E. Dry, Appl. Catal., A 138 (1996) 319–344.
[5] R.L. Espinoza, A.P. Steynberg, B. Jager, A.C. Vosloo, Appl. Catal., A 186 (1999)
13–26.
5. Conclusions
[6] K. Yokota, K. Fujimoto, Fuel 68 (1989) 255–256.
[7] S. Yan, L. Fan, Z. Zhang, J. Zhou, K. Fujimoto, Appl. Catal., A 171 (1998) 247–254.
[8] G. Jacobs, K. Chaudhari, D. Sparks, Y. Zhang, B. Shi, R. Spicer, T.K. Das, J. Li, B.H.
Davis, Fuel 82 (2003) 1251–1260.
[9] X. Huang, C.B. Roberts, Fuel Process. Technol. 83 (2003) 81–99.
[10] N.O. Elbashir, P. Dutta, A. Manivannan, M.S. Seehra, C.B. Roberts, Appl. Catal., A
285 (2005) 169–180.
[11] H. Tang, H. Liu, X. Yang, Y. Li, J. Chem. Eng. Chin. 22 (2008) 259.
[12] E. Durham, S. Zhang, Christopher B. Roberts, AIChE Annual Meeting (2008):
Presentation 678d.
[13] K. Yokota, Y. Hanakata, K. Fujimoto, Fuel 70 (1991) 989–994.
[14] X. Lang, A. Akgerman, D.B. Bukur, Ind. Eng. Chem. Res. 34 (1995) 72–77.
[15] D.B. Bukur, X. Lang, A. Akgerman, Z. Feng, Ind. Eng. Chem. Res. 36 (1997)
2580–2587.
[16] M. Claeys, E. van Steen, In: Andre Steynberg, Mark Dry (Eds.), Studies in Surface
Science and Catalysis 152 – Fischer Tropsch Technology, Elsevier Science, 2004,
pp. 601–680.
[17] P. Enrique Iglesia, Catal. A 161 (1997) 59–78.
[18] P. Senzi Li, S. Krishnamoorthy, A. Li, G.D. Meitzner, E. Iglesia, J. Catal. 206 (2002)
202–217.
We have used an iron-based FT catalyst (1 Fe:0.1 Zn:0.02 K:0.01
Cu by mol) for Fischer Tropsch Synthesis in a supercritical (SC-FTS),
gas phase (GP-FTS), and slurry-phase (SP-FTS) environment. In SC-
FTS, a significant selectivity in the diesel range to aldehydes and
methyl-ketones was observed (both products confirmed by GC-
MS). These products were not detected under GP-FTS or SP-FTS
operation. Additionally, another iron-based FT catalyst (1 Fe:0.01
Cu:0.02 K by mol) was studied under SC-FTS conditions and simi-
lar products were observed. We attribute this to the supercritical
media having a high capacity to extract and stabilize products,
inhibiting secondary reactions. Decreasing the residence time of
the reaction (decreasing conversion to make the product spec-
trum more primary) gave increased aldehyde and decreased olefin
selectivity. From this, we conclude that the aldehydes are primary
products and are converted by secondary reactions to olefins. Dop-
ing the supercritical media with octyl aldehyde resulted in the
aldehyde being incorporated into heavier products, indicating that
the process that forms the aldehyde is part of the propagation
mechanism. From this, we have concluded that CO Insertion is a
likely mechanism for the FTS reaction on iron-based catalysts.
[19] A. de Klerk, Fischer Tropsch Refining. PhD Thesis, University of Pretoria, Preto-
ria, South Africa, 2008.
[20] B.H. Davis, Catal. Today 141 (2009) 25–33.
[21] W. Keith Hall, R.J. Kokes, P.H. Emmett, J. Am. Chem. Soc. 82 (1960) 1027–1037.
[22] I. Wender, H.W. Sternberg, R.A. Friedel, S.J. Metlin, R.E. Markby, Technical
Report (1962) OSTI ID: 7193411.
[23] Y. Wang, B.H. Davis, Appl. Catal., A 180 (1999) 277–285.
[24] M. Zhuo, K. Fei Tan, A. Borgna, M. Saeys, J. Phys. Chem. C 113 (2009) 8357–8365.
[25] G. Blyholder, M. Lawless, Langmuir 7 (1991) 140–141.
[26] O.R. Inderwildi, S.J. Jenkins, D.A. King, J. Phys. Chem. C 112 (2008) 1305–1307.
[27] X. Dai, C. Yu, J. Nat. Gas Chem. 17 (2008) 365–368.
[28] N.O. Elbashir, C.B. Roberts, Ind. Eng. Chem. Res. 44 (2005) 505–521.
[29] M.E. Dry, Appl Catal., A 138 (1996) 319–344.
Acknowledgements
The authors would like to gratefully acknowledge financial
support from the U.S. Department of Energy National Energy Tech-
nology Lab and the U.S. Army TACOM through the Consortium for