Novel non-aqueous Fe(III)/Fe(II) redox couple for the catalytic oxidation of hydrogen sulfide to sulfur by air

Article abstract of DOI:10.1039/b513384j

A series of Fe(iii) salts and organic solvents have been screened to develop novel non-aqueous catalysts for the conversion of H₂S to sulfur. FeCl₃/95% N-methylpyrrolidinone/5% H₂O was found to be a most efficient non-aqueous system. The process chemistry, the proposed mechanism of catalytic oxidation of H₂S, and the electrochemistry are discussed. The Royal Society of Chemistry 2006.

Full text of DOI:10.1039/b513384j

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
Novel non-aqueous Fe(III)/Fe(II) redox couple for the catalytic
oxidation of hydrogen sulfide to sulfur by air
a
a
b
a
a
Guoxiong Hua, Qingzhi Zhang, Derek McManus, Alexandra M. Z. Slawin and J. Derek Woollins*
Received 21st September 2005, Accepted 7th December 2005
First published as an Advance Article on the web 19th December 2005
DOI: 10.1039/b513384j
A series of Fe(III) salts and organic solvents have been screened to develop novel non-aqueous catalysts
for the conversion of H
most efficient non-aqueous system. The process chemistry, the proposed mechanism of catalytic
oxidation of H S, and the CV are discussed.
2
S to sulfur. FeCl
3
/95% N-methylpyrrolidinone/5% H O was found to be a
2
2
Introduction
chelate of acetylacetone (2,4-pentanedione) in NMP, water free
or with a water content of less than or equal to 5 w/v%, was the
most successful catalyst system. The substantially non-aqueous
process was claimed to desulfurise, decarbonate (i.e. remove or
absorb carbon dioxide) and dehydrate a gas stream simultaneously.
However, a key deficiency in this system is found to be substantial
loss of ligand, b-diketone, that eventually led to deactivation of
Iron complex catalytic solutions for removing hydrogen sulfide
contained in a wide range of industrial gas streams have been used
for over a century and a half as redox catalysts. Current systems
to use aqueous aminopolycarboxylate chelate iron redox catalyst
solutions were introduced around 1962. Since then, numerous
improvements in process equipment, plant configuration and
process chemistry have been made. Actually, the treated gas
streams containing hydrogen sulfide are usually contaminated with
carbon dioxide (CO ) or other contaminating gases which still
need to be cleaned out from gas streams. The currently applied
Fe–NTA (nitrilotriacetic acid) system consists of two separate
processes, which use an aqueous catalyst system to convert H
to S and a non-aqueous phase to absorb CO , H O and other
organic contaminants. That is, CO , H O and other organic
contaminants are absorbed using an organic solvent such as
N-methylpyrrolidinone (NMP), H S is catalytically converted to
1
4
the catalyst system in the process.
2
Accordingly, attempts to further improve current non-aqueous
catalyst systems by selecting more robust ligands in place of
easily degraded b-diketones and to seek suitable organic solvents
were our initial objective. However, it is rather difficult to find a
suitable iron complex which is both soluble in organic solvents
and comparable to the iron complex of b-diketone in price and
practicality under the required conditions (pH = 8.5). We decided
2
2
S
2
2
2
2
to test if it is possible to oxidize H S to S directly with simple
2
iron(III) salts in an organic solvent. The simplest idea provided us
with surprising and excellent results. The preliminary experiment
2
sulfur by using an aqueous catalyst solution of Fe(III) complex.
Considering the economical and practicable reasons, it is
necessary to develop a more efficient process that can remove
not only sulfur contaminants, but also carbon dioxide and water
in a single process while producing or recovering a high quality of
solid elemental sulfur. A non-aqueous catalyst system could be a
unique possibility to solve the above problems in a single process
because a specific organic solvent employed, for example NMP,
could also absorb all the contaminant gas including water.
However, most of the conventional aqueous catalysts such as the
Fe(II) and Fe(III) complexes of NTA, EDTA (ethylenedinitrilote-
tracetic acid) and HEDTA (N-(2-hydroxyethyl)ethylenediamine-
with the FeCl –NMP catalyst system without any ligand did work
3
unexpectedly well for the conversion of hydrogen sulfide to sulfur.
This triggered further investigation of different iron salts in various
organic solvents.
Results and discussion
Selection of organic solvents
About twenty common organic solvents were screened based upon
the solubility and stability of FeCl
or with 5 w/v% of water) without pH control. If FeCl
3
in the solvent (either water-free
dissolves
3
ꢀ
ꢀ
N,N ,N -triacetic acid) are not adequately soluble in organic
solvents, nor are that of the phosphorus-containing analogues
well in a solvent to give a stable homogeneous solution (water-
free or with 5 w/v.% of water), the solution was then loaded on
the 1-litre reactor for preliminarily tests on catalytic activity for
of NTA or EDTA. Up to now, only the b-diketone (R
1
–CO–
CH –CO–R ) chelate complexes are feasible as catalysts in non-
2
2
the oxidation of H
2
S to sulfur by air. The catalytic experiments
aqueous processes for the removal of hydrogen sulfide from
natural gas. The metals patented as b-diketone chelates for this
purpose include polyvalent iron, copper, cobalt, vanadium, nickel
3
−1
were performed under conditions of H
2
S = 2.0 cm min ,
air = 700 cm min , circulation rate of absorption solution =
00 cm min , initial pH value and room temperature. Table 1 lists
3
−1
3
−1
1
3
and manganese, as well as scandium, yttrium or the elements
the qualitatively tested results of different FeCl
systems based on visual observations.
3
/organic solvent
in the lanthanide and actinide groups. Among them the Fe(III)
Furan, sulfolane, trichloroethylene, morpholine and N-
a
formylmorpholine (water-free or with 5 w/v% of water) do not
School of Chemistry, University St Andrews, St Andrews, Fife, UK KY16
9
ST. E-mail: jdw3@st-and.ac.uk
dissolve FeCl
further.
3
properly. Therefore these solvents were not tested
b
US Filter Corporation, Inc., Plainfield, IL, 60544, USA
This journal is © The Royal Society of Chemistry 2006
Dalton Trans., 2006, 1147–1156 | 1147

View Article Online
7
Table 2 Properties of N-pyrrolidinone (NMP)
Property
◦
High boiling point/ C
202
◦
◦
Low freezing point/ C
−24 (−12 F)
−
−
1
1
◦
−1
Low specific heat/cal g
Low toxicity/g (kg rat)
TSCA listed
C
0.4
LD50.7
Yes
Biodegradable
Low viscosity/cP
High flash point/ C
High auto ignition temperature/ C
Complete miscibility with water
Yes
◦
1.65 (at 25 C)
95 (204 F)
346 (655 F)
Yes
◦
◦
◦
◦
−
1
Inexpensive/£ lb
0.94
Assured availability
In-process stability
Dissolve sulfur
Yes, from 1,4-butanediol
Early work is favourable
Readily
0.4% at 20 C; 9% at 130 C
48 L/L NMP at 20 C, 1 atm
◦
◦
Steep sulfur solubility gradient
◦
High H
Good O
2
S solubility
solubility
2
S over CO selectivity
3
◦
2
55 cm /L NMP at 20 C, 1 atm
Yes
High H
2
FeCl
3
is soluble in a variety of other organic solvents, either
water-free or with 5 w/v% of water, to give stable homogeneous
solutions. The solvents include formamide (FA), N-methylpyrroli-
dinone (NMP), dimethyl sulfoxide (DMSO), dimethyl for-
mamide (DMF), propylene carbonate (PC), 1,4-dioxane (DO),
diethylglycolmonomethyl ether (DEGE), di-(ethylglyco)-diethyl
ether (DEGDE), triethylphosphate (TEPP), methyl-acetoacetate
(
MAA), triethanolamine (TEA), 2-butoxyethanol (BE), ac-
etamide and ethanol. However, as displayed in Table 1, only
the FeCl /NMP system, our initial discovery, was found to have
catalytic activity for the oxidation of hydrogen sulfide to sulfur by
air. FeCl in other organic solvents showed little catalytic activity.
3
3
Therefore, NMP became our solvent of choice for further studies.
Actually, N-methylpyrrolidinone (NMP) has been widely ap-
5
plied in industrial scale gas purification, for example in the
ꢀ
R
Purisol process for the physical removal of H
2
S and CO from
2
gas streams. In addition, it has also been widely used to remove
adhesives (especial cyanoacrylate-based) from plastic, glass, metal,
ceramic, stone and fiber surfaces and also from human skin and to
6
clear surfaces of plastic, especially poly(lauryl methacrylate) and
7
to dehydrate. As can be seen from Table 2, the inexpensive and
environmentally acceptable NMP can be a perfect organic media
for the conversion of hydrogen sulfide to sulfur with its ideal high
boiling point, low freezing point, low toxicity, high H
solubility and complete miscibility with water.
2
S and CO
2
Selection of ferric salts
Since FeCl
the oxidation of H
3
in NMP exhibits efficient catalytic activity toward
S to S, other iron salts may also have
2
similar activity in the same solvent for the same conversion.
Therefore, a number of ferric salts were tried in water-free
NMP or 95% NMP/5% water in the same procedure. The
ferric salts used in this study include ferric citrate, ferric ox-
alate (Fe
tonate( [CH
2
(C
2
O
4
)
3
), ferric carbonate (Fe
2
(CO ) ), ferric acetylace-
3
3
−
C(O)CH=C(O )CH
]
Fe), ferric hydroxyl acetate
(SO O), ferric nitrate
·5H
. The results are listed in Table 3
3
3
3
(
(
Fe(OH)(CH
3
CO
2
)
2
), ferric sulfate (Fe
PO
2
4
)
3
2
Fe(NO
3
)
3
·9H
2
O) and Fe(H
2
2 3
)
and compared with that of FeCl .
3
1
148 | Dalton Trans., 2006, 1147–1156
This journal is © The Royal Society of Chemistry 2006

View Article Online
Table 3 Screening of ferric salts in 95% NMP/5% water for the catalytic oxidation of hydrogen sulfide by air
Concentration of
Fe(III)/mmol L
Test
time/h
−
1
Ferric salt
Results and comments
Iron(III) acetylacetonate,
90
36
5
Iron salt partly dissolved to give a suspension, dark solid formed, trace of tail gas
observed and poor catalytic activity
Fe(OH)(OAc)
2
78.5
Iron salt partly dissolved to give a suspension, no tail gas observed, grey S
8
(3.727 g
mixed with unknown dark solid) and residue of S 4.70 g in the filtration (analysis by
HPLC), severe loss of iron ion: Fe ion concentration changed from 2084 to 944 mg L
and not ideal catalytic system
8
−
1
Fe(H
2
PO
2
)
3
90
3
Iron salt partly dissolved to give a suspension, dark solid formed, severe tail gas and poor
catalytic activity
a
2−
2−
4
Ferric nitrate (Fe(NO
3
3
)
)
3
3
)
)
90
24
Low yield of S
8
: 63.03%, high conversion to SO
3
(28.56%) and to SO
(1.05%), severe
tail gas in the process and poor catalytic activity
Severe tail gas found in 9.5 hours, high ratio of Fe /Fe : 94.42% and poor catalytic
b
2+
3+
Ferric nitrate (Fe(NO
90
10.5
80.5
activity
c
2−
Ferric sulfate (Fe
2
(SO
4
)
3
)
90
Moderate yield of S
.92%, low conversion to SO
activity
Severe tail gas in 10 hours, low yield of S
8
: 88.50%, severe tail gas: CuS 10.77%, low conversion to SO
3
:
2−
−1
−1
1
4
, high loss of iron: 34.21 mg L
h
and poor catalytic
d
Ferric sulfate (Fe
Ferric sulfate (Fe
2
2
(SO
(SO
4
)
)
3
3
)
)
90
14
8
: 48.49%, unknown water-soluble precipitate,
and poor catalytic activity
−
1
−1
severe loss of iron: 288 mg L
Severe tail gas in 9 hours, high ratio of Fe /Fe (92.33%) and poor catalytic activity
High quality of S
low yield of SO
h
e
2+
3+
4
90c
9.5
300
2−
Ferric chloride (FeCl
3
)
90
8
−
: yellow crystalline, high yield of S
8
: 98.72%, low yield of SO
3
: 0.02%,
2
4
: 0.09%, no iron loss, no tail gas and perfect catalytic system
a
b
c
Initial pH value without any pH control during the process. pH controlled to 1.00 during the process. Initial pH value and does not control pH during
d
the process. The pH of the starting solution was adjusted to the same as that of 90 mmol FeCl
the process. pH controlled to 1.00 by addition of 2 N of NaOH during the process.
3
/95% NMP/5% water. No further control of pH during
e
2
−
2−
4
Ferric acetylacetonate, Fe(H
2
PO
2
)
3
and ferric hydroxyl acetate
for sulfur (98.7%), low conversion to S
2
O
3
(0.02%) and SO
(
Fe(OH)(CH CO ) are partly soluble in 95% NMP/5% water
3
2 2
)
(0.09%). No tail gas was observed after 300 hours of running.
to give a slight suspension. The suspensions showed low catalytic
activities when employed as catalysts for the oxidation of hydrogen
sulfide to sulfur. The same unsatisfactory results are obtained when
other organic solvents were used to substitute for pure NMP or
Influence of concentration of FeCl
3
on catalytic activity
While other organic solvents and iron salts show negative or
9
5% NMP/5% H
Both ferric sulfate and ferric nitrate dissolve in pure NMP
2
O.
poor results for the catalytic oxidation of H S to S by air, the
combination of FeCl with NMP provides a perfect catalyst for
this conversion. Table 4 shows the influence of FeCl concentration
on the catalytic activity in pure NMP and 95% NMP/5% water.
The maximum solubility of FeCl in pure NMP or 95%
NMP/5% H O was determined. Also tested was FeCl O.
·6H
Both anhydrous FeCl and FeCl O mixed with pure NMP
2
3
very slowly. In 95% NMP/5% water, however, the two salts
dissolved readily to give a stable solution with concentrations
up to 360 mmol L . The pH value of the resulting solution,
lower than 1, was slightly different between the two salts and
the various concentrations. Catalytic experiments were performed
3
−
1
3
2
3
2
3
3
·6H
2
with samples of 90 mmol of Fe
water and 90 mmol of Fe(NO
2
(SO
·9H
4
)
3
·5H
2
O in 95% NMP/5%
or 95% NMP/5% water to give a stable solution, although
the latter dissolved in pure NMP rather slowly. The maximum
3
)
3
2
O in 95% NMP/5% water
−
1
with both uncontrolled pH or controlled pH (to 1 by addition
of 2 N NaOH). Both salts displayed catalytic activities for the
catalytic oxidation of hydrogen sulfide to sulfur to some extent.
However, their catalytic activities were found to decrease gradually
with an increase in the running time. The catalysts were completely
deactivated eventually. A high proportion of Fe /Fe (more
than 90%) was found in the resultant solution for both catalyst
systems. Obviously, the Fe formed during the process is hard
to re-oxidize to Fe . In both instances, the tail gas became
gradually more severe with running of the experiment. In one
case of 90 mmol of Fe
whose pH value has been previously adjusted to the same as
that of the corresponding FeCl solution with 2 N H SO , an
unknown water-soluble solid was found to precipitate out. The
concentration of iron ion was reduced dramatically during the
process.
concentration of FeCl
3
−
anhydrous is 270 mmol L in pure
1
NMP and 180 mmol L in 95% NMP/5% water. While that of
−
1
FeCl
3
·6H
2
O in 95% NMP/5% water is 180 mmol L . Experiments
were carried out with concentrations of FeCl
of 18 to 90 mmol.
3
varying in the range
2
+
3+
When 18 mmol of FeCl
3
in pure NMP was tested for the
conversion of H S to sulfur by air, no tail gas was observed within
2
2
+
the first 17 hours. Longer running of the experiment led to the
disclosure of trace tail gas. This can be explained by the shortage
3
+
3
+
of the Fe ion in the process. In the initial stages of the experiment,
3
+
2
(SO
4
)
3
·5H
2
O in 95% NMP/5% water
there was enough Fe in the system to oxidize hydrogen sulfide.
3
+
2+
As the redox process continued, the Fe and Fe ions reached
3
+
3
2
4
equilibrium, at which point the content of Fe was insufficient to
convert hydrogen sulfide completely to sulfur, as a consequence,
the excess of hydrogen sulfide was released in the tail gas. Even so,
the absorption of hydrogen sulfide was still very impressive. The
2
−
In clear contrast, 90 mmol of ferric chloride in 95% NMP/5%
formation of relatively high concentrations of by-products SO
3
2
−
water showed excellent efficiency for the catalytic oxidation of H
2
S
and SO
4
might result from the direct oxidation of H S by the
2
to sulfur by air. The system works perfectly with high selectivity
oxygen (air) in the oxidizer zone.
This journal is © The Royal Society of Chemistry 2006
Dalton Trans., 2006, 1147–1156 | 1149

View Article Online
Table 4 Catalytic activity of FeCl
3
/NMP or 95% NMP/5% water system
100% NMP 100% NMP
Solvent system
Ferric salt
100% NMP
95% NMP/5% H
2
O
2
95% NMP/5% H O
FeCl
18
3
FeCl
36
300
3
FeCl
90
313
3
FeCl
90
300
3
FeCl
90
198
3
·6H
2
O
−
1
Iron salt/mmol L
Test time/h
100
Crystalline S
Residual S in filtrate/g
Tail gas (CuS)/g
8
/g
11.846
0.226
0.184
4.91
34.010
1.077
0.135
4.12
37.780
0.819
0
4.56
11.03
16.06
95.84
0
0.73
0.88
2.56
5033.55
5035.91
0
46.968
7.167
0
0.20
1.50
4.13
98.72
0
0.02
0.09
19.296
3.491
0
3.00
2.89
1.00
98.49
0
0.83
0.40
8
2−
C
C
SO3 in the resulting solution/mM
2−
SO4 in the resulting solution/mM
S in the resulting solution/M
3.66
4.34
Residual H
2
13.33
90.32
0.46
2.36
0.46
18.50
95.15
0.12
0.72
0.12
a
Conversion of H
Conversion of H
Conversion of H
Conversion of H
2
2
2
2
S to S
S to CuS (%)
S to SO
S to SO4 (%)
8
(%)
a
2
−a
3
(%)
2
−a
Residual H
2
S (%)
6.40
3.26
1.17
0.28
−
−
1
Iron content of the starting solution/mg L
Iron content of the resulting solution/mg L
1002.21
1004.95
0
1999.55
2006.10
0
5042.57
5104.17
50.06
42.37
44.86
5152.75
5279.79
49.65
43.13
44.05
1
−
1
H
H
2
O content of the starting solution/g L
O content of the resulting solution/g L
−
1
2
2
+
3+
Fe /Fe in the process (%)
40.20
45.55
50.10
a
Calculation based on the amount of all components containing sulfur, which is supposed to be 100%.
Doubling the concentration of FeCl
led to the enhancement of selectivity to sulfur from 90.3% to
3
to 36 mmol in pure NMP
that in the starting solution. This may arise from the evaporation
of H O in the process, leading to the slightly condensed reactant
solution. The degradation of NMP, which will be discussed below
in detail, may also be responsible for the condensation of solvent
during the process.
2
2
−
9
5.2%. The concentration of the over-oxidized components SO
3
2
−
and SO
4
decreased. The loss of H S in the tail gas (0.12%),
2
though lower than in the case of 18 mmol (0.46%), was still ob-
3
+
served, suggesting that the Fe ion in the equilibrium was still not
abundant. An increase in FeCl concentration to 90 mmol in 100%
of NMP stopped the loss of H S in the tail gas completely in spite of
the fact that the yield of sulfur was not further improved (95.84%).
3
Effect of the anion of the ferric salt in water
2
Due to the drastic different effects of various iron salts on the
2
−
A slight increase in the side product SO
4
implied that the
catalytic activity for the conversion of H
to investigate the same iron salts in water for the same conversion.
Surprisingly, the aqueous FeCl , Fe (SO and Fe(NO show
extreme similarity to each other with very low catalytic activity
2
S to S in NMP, we decided
concentration of iron ion might be a little bit higher than required.
Further improvement for the selectivity to sulfur was achieved
by a slight modification of the solvent with the addition of 5 w/v%
of water. As shown in Table 4, 90 mmol L ferric chloride in
5% NMP/5% water afforded 98.7% yield of sulfur, while the
3
2
4
)
3
)
3 3
−
1
3+
2+
(Table 5). Actually, this similar irreversibility of Fe to Fe has
been reported before. When FeCl or Fe (SO or Fe(NO are
9
3
2
4
)
3
)
3 3
2
−
2−
concentration of side products SO
3
and SO
O in NMP somewhat
reduced the solubility or residential time of H S in the absorption
solution and might have increased the rate of the conversion of
S to sulfur. There was less residual H S in 95% of NMP than in
water-free NMP.
0 mmol of FeCl
similar results to its anhydrous counterpart in spite of the slight
increase of H O content in the reaction solution. However, if the
4
slightly decreased.
dissolved in water, a series of hydroxylated ferric ions, which form
by hydrolysis of the Fe(III) ion, have been found to dominate in
Actually, the presence of 5 w/v% of H
2
8–16
2
solution. When a stirred cell reactor for absorption of hydrogen
sulfide into aqueous solutions of ferric sulfate and ferric chloride
respectively in the absence of any organic ligands, the hydroxylated
H
2
2
2
+
ferric ion [Fe(H
2
O) OH] was found to be the actual reactive
5
17
2+
9
3
·6H
2
O in 95% NMP/5% water showed closely
species in the processes, and the hydrated ferrous ion [Fe(H
2
O)
6
]
2
+
was the major form of Fe in the solution which was found to be
3
+
2
hard to oxidize to Fe in air. That is, in the aqueous solution,
the hydrated Fe is very stable and difficult to oxidize to hydrated
or hydroxylated Fe ion. As a consequence, the influence of the
anion in the iron salt on the aqueous catalytic activity, which could
be ignored in practice, is far lower than that of the solvent water
molecule itself.
2
+
content of water in the catalyst system reached 20%, the catalytic
activity was found to reduce sharply—severe tail gas re-appeared
in the process.
All the experiments listed in Table 4 gave high quality bright
yellow crystalline sulfur, which is in clear contrast to the grey sulfur
contaminated with FeS from the Fe–NTA system. In addition to
the crystalline sulfur, there is still a considerable amount of sulfur
3
+
The possible catalytic mechanism of iron salts in NMP
dissolved in the solution. Work-up by addition of H O precipitated
2
the sulfur out as a yellow solid, also in high quality. The solvent
can be recycled by evaporation of the excess water.
On changing the solvent from water to pure NMP or 95%
NMP/5% H O, the effect of the anion on the catalytic activity of
2
No iron loss was observed in the process. For the catalyst system
the iron salt becomes vital, implying that the reactive species in the
catalytic cycle must be an anion-containing complex. This is not
difficult to understand. The NMP is far less polar than water. Iron
with 90 mmol FeCl
3
·6H
2
O or FeCl in 95% of NMP, the total iron
3
concentration in the resulting solution was a little bit higher than
1
150 | Dalton Trans., 2006, 1147–1156
This journal is © The Royal Society of Chemistry 2006

View Article Online
2
Table 5 Catalytic activities of the various ferric salts in H O
Testing
time/min
Catalyst composition
FeCl
([Fe] = 90 mM)/100% H
Testing conditions
Results and comments
3
−1
3
2
O
H
7
2
S: 2.00 cm min ; air =
700
555
630
pH 0.44 changed to resulting pH 0.20, no tail gas found in
the first 600 min, conversion of H S to S (58.03%), CuS:
3
−1
00 cm min ; without any
2
8
3+
2+
control of pH during the process
0.29 g and no iron ion loss, high Fe /Fe in the resulting
solution: 95.51%, and positive activity observed within the
first 600 min, but it gradually dwindled away due to hard
2+
3+
conversion of Fe to Fe during the process
3
−1
Fe
2
(SO
4
)
3
([Fe] = 90 mM)/100% H
2
O
H
7
2
S: 2.00 cm min ; air:
Initial pH = 1.07, no tail gas found in the first 550 min,
3
−1
00 cm min ; control pH of
solution to 1.00 by addition of
2 8
conversion of H S to S (1.18 g, 80%), CuS (0.07 g) and no
2+ 3+
iron ion loss, high Fe /Fe in the resulting solution:
92.33%, and positive activity was observed in the first
2
N NaOH during the process
6
30 min, but it gradually dwindled away due to hard
2+ 3+
conversion of Fe to Fe
Initial pH = 0.95, no tail gas found in the first 570 min,
conversion of H S to S (1.36 g, 81%), CuS (0.07 g) and no
iron ion loss, high Fe /Fe in the resulting solution:
94.42%, positive activity observed within the first 630 min,
but it gradually dwindled away due to hard conversion of
3
−1
Fe(NO
3
)
3
([Fe] = 90 mM)/100% H
2
O
H
7
2
S: 2.00 cm min ; air:
3
−1
00 cm min ; pH of the
2
8
2+
3+
solution controlled to 1.00 by
addition of 2 N NaOH during
the process
2+
3+
Fe to Fe in the process
salts thus have less possibility for ionization in NMP. Therefore the
solution is not dominated by totally solvated iron as in the aqueous
system. The complexes FeCl
3
or FeCl
3
·6H
2
O, Fe(NO
3
)
3
·9H
2
O and
Fe (SO O in NMP or 95% NMP/5% H
2
4
)
3
·5H
2
2
O can be rep-
(3−x)+
(3−x)+
y x
Z ]
resented as [Fe(NMP)6−xCl
x
]
, [Fe(NMP)6−x−y(H
2
O)
(H O)
−
−
(2−2x)+
(
Z = Cl , NO
3
) and ca. [Fe(NMP)5−2x−y(SO
4
)
x
2
y
]
,
respectively. The following are the mechanisms we have tentatively
proposed for the iron catalytic conversion of H
in NMP.
2
S to sulfur by air
As shown in Schemes 1–4, iron salts in NMP or in 95%
3
+
NMP/5% H O could initially form partly solvated Fe ions or
2
complexes [step (1)]. Hydrogen sulfide then dissolves in a solution
containing the iron salt and NMP to afford new complexes
Scheme 2 A suggested catalytic cycle in the non-aqueous catalyst system
of FeCl or FeCl O/NMP or 95% NMP/5% H O for the conversion
·6H
of H S to S.
3
3
2
2
2
containing sulfur by ligand exchange or substitution reaction in
−
−
2−
steps (2)–(4) in which the anion ion such as Cl or NO
3
or SO
4
is
−
substituted by HS . One NMP molecule may be displaced during
3
+
the process. Electron transfer between the sulfide and Fe within
the sulfur-containing Fe complex takes place in step (5) and step
(6). The free radical of sulfur-containing complex, which is formed
Scheme 1 A suggested catalytic cycle in the non-aqueous catalyst system
of FeCl /NMP for the conversion of H S to S.
3 2
This journal is © The Royal Society of Chemistry 2006
Dalton Trans., 2006, 1147–1156 | 1151

View Article Online
energy. Addition of the appropriate amount of water (5%) to the
FeCl
3
/NMP system reduces the residence time of H S in solution
2
to some extent and decreases the possibility for further oxidation,
and therefore is beneficial to the selectivity to sulfur. A limited
number of H O molecules may also be coordinated to the iron in
2
the reactive species as shown in Schemes 2, 3 and 4. The active
energy in Scheme 2, which contains water in step (2), may be lower
than that in Scheme 1, so the formation of sulfur is quicker in
the presence of water. However, if the water content reaches 20%,
water molecules would compete with NMP or the anion ligands to
2
+
co-ordinate with iron, especially with Fe to give totally hydrated
iron, which as in the aqueous system, loses its catalytic activity
sharply.
In Schemes 3 and 4 water is considered as part of the coordinated
ligands, step (7) is vital for the whole cycle as suggested by
2
+
3+
the high proportion of Fe /Fe (up to 94.4%) in the resulting
2
+
solution. Like the hydrated Fe in the aqueous system, the nitrate
or sulfate-containing Fe(II) species in Schemes 3 and 4 are quite
stable and very difficult to oxidize into their corresponding Fe(III)
counterparts with air. Therefore, the catalytic cycles were blocked
in step (7) for the Fe
2
(SO
4
)
3
/NMP and Fe(NO
the diminishing Fe in the system directly leading to the gradual
loss of catalytic activity of Fe (SO and Fe(NO in NMP for
the conversion of H S to S with air.
3
) /NMP systems,
3
3
+
2
4
)
3
)
3 3
2
How can the stability of the Fe(II) species of chloride, nitrate and
sulfate or hydrogen sulfate be so different? Firstly, according to the
Scheme 3 A suggested catalytic cycle in the non-aqueous catalyst system
of Fe(NO O/95% NMP/5% H O for the conversion of H S to S.
·9H
3
)
3
2
2
2
−
−
−
2−
Lewis theory of acids and bases, Cl , NO
3
, HSO
4
and SO
, H SO
, respectively. The acidity of the above acids and
4
are the corresponding conjugate bases of HCl, HNO
and HSO
3
2
4
−
4
18
basicity of the conjugate bases are found to follow the order shown
below:
H
2
SO
4
(pK
HNO
a
= −9) > HCl (pK
(pK
a
= −7) >
−
3
a
= −1.4) > HSO
4
Increasing acid strength ←
2
−
−
−
−
4
SO
4
> Cl > NO
3
> HSO
Increasing base strength ←
Cl is a typical weak monodentate ligand while SO
−
2−
−
3
4
and NO
can be regarded as weak bidentate ligands due to their special
resonance structures. Therefore it is reasonable that iron chloride
species are better solvated or replaced by solvent than iron nitrate
3
+
and iron sulfate species in NMP. Furthermore, Fe is generally
believed to prefer to coordinate nitrogen-containing ligands
rather than oxygen-containing ligands to form stable complexes.
2
+
On the contrary, Fe tends to coordinate oxygen-containing
rather than nitrogen-containing ligands. Therefore, the more
N-donor ligand NMP molecules coordinated to the iron ion,
Scheme 4 A suggested catalytic cycle in the non-aqueous catalyst system
of Fe (SO O/95% NMP/5% H O for the conversion of H S to S.
·5H
2
4
)
3
2
2
2
3
+
the easier the Fe regeneration and the more active the iron
species is for catalysis. A similar point of view has also been
in step (5), readily decomposes to release the elemental sulfur, and
2
+
18
2+
to dissociate a Fe solvated ion in step (6). A regeneration of the
catalyst is carried out in step (6) and step (7).
previously reported before. We believed that the Fe species
in the Fe(NO /NMP or Fe (SO /NMP systems must be
stabilized by oxygen-donor sulfate or nitrate ligands and this has
3
)
3
2
)
4 3
In Scheme 1 or Scheme 2, step (2) can be rate determining,
because the step that releases a free radical needs higher activation
3
+
proved to be very difficult to oxidize to Fe with air.
1
152 | Dalton Trans., 2006, 1147–1156
This journal is © The Royal Society of Chemistry 2006

View Article Online
Table 6 Half-wave potentials (V vs. saturated calomel electrode, SCE) ([Fe] = 0.02 M, initial pH value)
System FeCl /95% NMP/5% H Fe(NO /95% NMP/5% H
3
2
O
3
)
3
2
O
Fe
2
(SO
4
)
3
/95% NMP/5% H
2
O
E
1/2/V vs. SCE
−0.393
−0.307
−0.066
3
+
2+
Cyclic voltrammogram study of non-aqueous catalyst systems
In order to further explore the difference in the catalytic activi-
For the Fe /Fe redox couple, the standard potential E for
0
the reduction of the ferric ion to the ferrous ion is +0.770 V under
standard conditions. As shown in Table 6, the half wave potentials
ties between Fe(NO /NMP, Fe (SO /NMP and FeCl /NMP
3
)
3
2
4
)
3
3
3+
2+
E
1/2 of Fe /Fe in a non-aqueous system, which depends on
the nature of the anion ion, are all much more negative than
the standard potential E . Though it is still difficult to correlate
the wave potentials to the catalytic activity of the Fe /Fe
redox couple directly, there is indeed an obvious difference
between the wave potential of Fe /Fe of the catalytically active
FeCl /95% NMP/5% H O system and the poor catalytic system
of Fe(NO /95% NMP/5% H O and Fe (SO /95% NMP/5%
O. Further studies are needed to establish the relationship
for the catalytic oxidation of hydrogen sulfide to sulfur, their
CVs were measured. The cathode wave in Fig. 1–3 can be
assigned to the reduction process, and the anodic wave to a
complementary oxidation reaction. The half wave potentials of
0
3+
2+
FeCl
3
/95% NMP/5% H
2
O, Fe(NO
3
)
3
/95% NMP/5% H
2
O and
3+
2+
Fe
2
(SO
4
)
3
/95% NMP/5% H O are listed in Table 6.
2
3
2
3
)
3
2
2
)
4 3
H
2
3
+
2+
between the half wave potential E1/2 of Fe /Fe and the actual
catalytic activity.
Process chemistry
The process chemistry for the air oxidation of H
catalysed by the non-aqueous system is, actually, closely similar
to that in the aqueous systems. The H S is induced into a
2
S to sulfur
2
circulating catalyst solution in the absorber zone where it is
Fig.
1
Cyclic voltammogram of Fe(NO
3
)
3
/95% NMP/5%
H
2
O
solvated and physically removed, simultaneously, the solvated H S
reacts with the Fe(III) solvate forming elemental sulfur which
2
(
[Fe(III)] = 0.02 M, initial pH value).
does not precipitate but remains in the organic solvent—NMP
or 95% NMP/5% H O—in the early stages. The Fe(III) solvate
2
is concurrently reduced to the Fe(II) solvate state. Reactions
occurring in the absorber are represented as follows.
The sulfur produced dissolved continuously in the organic
solvent. Once the organic solvent is saturated with dissolved
sulfur, the yellow crystalline sulfur precipitates gradually in both
absorber zone and oxidizer zone. In industry, the whole process
◦
can be operated at higher temperatures (60–70 C) to increase the
solubility of sulfur. Crystalline sulfur product can be harvested by
cooling the reaction mixture.
H
2
S + Solvent ꢀ H
2
S (dissolved)
Fig.
2
2 4 3 2
Cyclic voltammogram of Fe (SO ) /95% NMP/5% H O
(
[Fe(III)] = 0.02 M, initial pH value).
H
2
S (dissolved) + 2 Fe(III) Solvate → S (dissolved) + 2 Fe(II)
The combined oxidizer–crystallizer provides sufficient residence
time for the regeneration of the active Fe(III) solvate and crystal-
lizaton of dissolved sulfur. The process can be represented by the
following equations.
O
2
(g) + Solvent ꢀO
2
(dissolved)
2
Fe(II) Solvate + 1/2 O (dissolved) → 2 Fe(III) Solvate
2
S (dissolved) → S (s)
The overall process can be represented by the following equation.
Fig.
3
Cyclic voltammogram of FeCl
3
/95% NMP/5%
2
H O
(
[Fe(III)] = 0.02 M, initial pH value).
H
2
S(g) + 1/2 O
2
(g) → S (s) + H
2
O
This journal is © The Royal Society of Chemistry 2006
Dalton Trans., 2006, 1147–1156 | 1153

View Article Online
7
Table 7 Advantages of non-aqueous FeCl
3
/NMP liquid redox sulfur recovery process versus the aqueous chelated iron redox sulfur recovery process
Consideration
Aqueous system
Novel non-aqueous system
FeCl /NMP/H
Needs no special ligand. The NMP functions as both
the organic solvent and organic ligand
Bright yellow crystalline sulfur is obtained by cooling
the circulating solution in certain zones. No solid
sulfur is deposited elsewhere. Sulfur can be washed
with neat solvent then water to recover valuable
chemicals. High efficiency gas spargers and packed
absorber towers can be used. It is very easy to get
crystalline product sulfur by filtration
Components of catalyst
Organic ligand
Fe(III) complex/stabilizer/H
2
O
3
2
O
Required a robust ligand. A key deficiency was the
rapid degradation of the ligand in the process
Grey amorphous precipitate, fouls equipment
especially gas/liquid contactor and gas spargers.
Precludes use of efficient porous metal spargers. It is
difficult to obtain the product sulfur by filtration
Sulfur product
Scavenger or inhibitor
A constant concentration of scavenger or inhibitor of
the hydroxyl radical or other free radicals was
required to prevent the degradation of ligand
No need
H
2
O removal
Produced sweet gas is saturated with water at about
Dehydration to pipe-line specification is achieved by
the hygroscopic solvent
◦
4
4 C. Dehydration requirements can be increased
Carbon dioxide removal
Only minimal CO
are needed if the sour gas contains more than 2% CO
2
removal is achieved Amine plants
Complete removal of CO
oxidized to sulfur, recovery of CO
2
is possible. As H
2
S is
2
2
as a resource is
possible. Market size for the process is expanded
considerably
Converts to crystalline sulfur
Carbonyl sulfide removal
Inert, could not remove
Presumed space requirements
Relative large, especially if an amine plant and a
glycol dehydration unit are also required
Considerably more compact and therefore cheaper.
Much more suited to smaller, remote wells.
Attractive for off shore platform applications. O
solubility in NMP is greater than in water. 48 L of
S dissolve in 1 L NMP
2
H
2
Corrosion
Stainless steel equipment required
Special materials equipment required because of the
acidity of the absorbed solution and the likely
−
corrosion of equipment by Cl
Surfactant application
Antifoam addition
Biocide addition
Needed to wet and sink sulfur
Occasionally required
Normally used
Required in initial catalyst fill, subsequent additions
are sometimes required
No need
Probably no need
Probably unnecessary
May not be needed
Chelate stabilizer addition
The processes of formation of the side products can be represented
by the following equations.
Degradation of NMP
NMP used in non-aqueous Fe(III)/Fe(II) redox processes for the
2
−
+
2
H
H
2
S + 3 O
2
→ 2 SO
3
+ 4 H
conversion of H S to S acts as both ligand and solvent, that is,
2
NMP involves both physical and chemical processes during the
absorption and conversion of H S to sulfur. Although under acidic
conditions (in contrast to the basic conditions used in the Fe
2
−
+
2
S + 2 O
2
→ SO
4
+ 2 H
2
chelate aqueous process), there is less opportunity for the free
hydroxyl radicals to survive, it can still be expected that NMP
may experience degradation to some extent. HPLC analysis did
confirm this slight decomposition of NMP.
Comparison with aqueous processes
As mentioned above, the absorbed solution also serves to remove
CO and water, if they are present in a real industrial process, by
2
absorption into the solvent. Hence the process can simultaneously
desulfurize, decarbonate and dehydrate sub-quality natural gas
in a single operation while giving high quality sulfur. This is in
marked contrast to the traditional combination of amine plant,
glycol dehydrator and sulfur recovery unit. The advantages of this
Table 8 shows the change in the organic components of
the FeCl
3
/NMP and FeCl
3
/95% NMP/5% H O systems after
2
running for ca. 300 hours as determined by HPLC. As mentioned
in Table 4, an appropriate amount (ca. 5%) of water is beneficial
to the selectivity to sulfur. However, water also accelerates the
oxidative degradation of NMP due to the increased opportunity
4
novel non-aqueous process, as indicated by Ferm, over aqueous
redox processes have been listed in Table 7.
of H
2
2
O formation.
Table 8 Results of HPLC analyses of the non-aqueous catalyst systems
FeCl
3
([Fe] = 90 mmol)/100%
FeCl
3
([Fe] = 90 mmol)/95% NMP/
Type of sample
Component
NMP (test time: 313 hours)
5% H O (test time: 300 hours)
2
Starting solution
Resulting solution
NMS
NMP
NMS
NMP
0
0
100%
2.78%
96.99%
100%
10.78%
88.77%
1
154 | Dalton Trans., 2006, 1147–1156
This journal is © The Royal Society of Chemistry 2006

View Article Online
Conclusions
potential in the absorber and oxidizer were not monitored. The
H
2
S inlet was controlled by a mass flow rate controller (MFC) and
air was gauged by a rotometer. The tail gas was passed through to
a CuSO (1.0 M) absorber. All experiments were accomplished at
room temperature.
A series of Fe(III) salts and organic solvents have been screened to
develop novel non-aqueous catalyst for the conversion of H
sulfur. FeCl /95% NMP/5% H O proved to be the most efficient
non-aqueous system.
The influence of concentration of iron ions has been investigated
simply. Satisfactory results were obtained with 90 mM of FeCl in
NMP or 95% NMP/5% H O.
2
S to
4
3
2
3
1000 cm of iron(III) salt in organic solvent or 95% organic
solvent/5% H O solution was placed in the oxidizer compartment.
2
3
The liquid pump was then started and part of the solution was
pumped into the absorber compartment with a flow rate of
2
NMP functions as both solvent and coordinating agent to
form a series of anion-containing solvated complexes of Fe(III)
3
−1
1
00 cm min . At this stage air was bubbled into the oxidizer
and the flow rate of the air was adjusted to the required value.
The valve of the H S cylinder was turned on and the H S flow
and Fe(II). In the case of FeCl
3
, due to the loose coordination
2
2
−
of the Cl ion to the Fe ion, it is believed that the Fe ion is
highly solvated; Fe(II) ions can be easily oxidized to Fe(III) ions
with air, so the catalytic cycle proceeds smoothly. In the cases
3
−1
rate was adjusted to 2.00 cm min . The time of the start of the
reaction was recorded. The produced sulfur was filtered off at
3
5
0 hour intervals and 5 cm of sample was taken from the running
of Fe
2
(SO
4
)
3
and Fe(NO ) , however, due to the higher affinity
3
3
system at set time intervals in order to monitor the changes in the
process, i.e. the change of iron ion concentration and the oxidative
degradation of organic solvent etc.
of oxygen donor sulfate and nitrate and their stabilizing effect
on Fe(II), the catalytic cycle is blocked at the regeneration step
of Fe(III) with air. Therefore, only FeCl
NMP/5% H O are catalytically active systems for the conversion
of H S to sulfur.
The influence of the anion of the Fe(III) salt is also reflected
in cyclic voltrammograms of the FeCl /95% NMP/5% H O,
Fe (SO /95% NMP/5% H O and Fe(NO /95% NMP/5%
O systems which were measured at initial pH value and [Fe] =
0 mM. FeCl /95% NMP/5% H O showed the most negative half
wave potential and was the most useful system as catalyst for the
removal of H S.
Briefly, 90 mM of FeCl
3
/NMP or FeCl
3
/95%
2
HPLC analyses of NMP and its related composition species
2
A Milton Roy Pump No.043 024 equipped with a chromatography
accessory, UV-visible spectromonitor 3000 (LDC analytical) and
a chart recorder (Water 746 Dato Module) were used.
3
2
2
4
)
3
2
)
3 3
H
2
Analytical column: Spherisorb ODS II 5 lm 4.6 mm × 250 mm;
2
3
2
column mobile phase; acetonitrile : water = 5 : 95; flow rate:
3
−1
1
.0 cm min ; temperature: ambient; detector: UV absorption at
2
2
20 nm; injection sample: 10 lL; recorder: 10 mV full scale; chart
3
or FeCl
3
·6H
2
O in 95% NMP/5% H O
2
−
1
speed: 0.5 cm min .
.04 g of the reaction mixture was taken from the oxidizer
compartment of the 1-liter continuous reactor and placed in a
proved to be a highly efficient catalyst for air oxidation of hydrogen
sulfide to sulfur to give high quality sulfur. However, in spite of the
fact that there are a lot of advantages to the FeCl /NMP catalyst
system over the aqueous iron complexes catalyst system, the
concern over the corrosive nature of Cl might limit its commercial
0
3
3
100-mL volumetric flask. This sample was diluted to 100 cm with
water and then filtered through a 0.45 lm micron filter before
analysis.
−
applications.
Quantitative analyses of NMP and its degradation intermedi-
ates were based on the calibration curve established by varying
the concentration of the corresponding known samples. Peak
heights at the indicated amplification settings were plotted vs.
concentration of the species. The linear nature of these plots
indicated precise HPLC determinations of these molecular species
over a considerable concentration range.
Experimental
Materials and agents
Unless otherwise stated, all chemicals were obtained from
Aldrich or Lancaster and were used directly without further
purification: formamide (FA), N-methylpyrrolidinone (NMP), N-
formylmorpholine (NFMP), morpholine (MP), dimethyl sulfox-
ide (DMSO), dimethyl formamide (DMF), propylene carbonate
Determination of water by Karl Fisher reagent
3
3
1
1
0 cm of Karl Fisher reagent A (1 cm ⇔3 mg of water) and
0 cm Karl Fisher reagent B (1 cm ⇔3 mg of water) were mixed
. The solution was then added dropwise to the sample
(
PC), 1,4-dioxane (DO), diethylglycolmonomethyl ether (DEGE),
di(ethylglyco)diethyl ether (DEGDE), triethylphosphate (TEPP),
,4-pentanedione (ACAC), triethanolamine (TEA), trichloroethy-
3
3
under dry N
2
2
while stirring until the colour of the solution changed from deep
brown to bright yellow.
lene (TCE), 2-butoxyethanol (BE), sulfolane, acetamide, furan,
ethanol, ferric citrate, ferric oxalate, ferric carbonate, ferric acety-
lacetonate, ferric acetate, ferric sulfate, ferric nitrate, iron(III) chlo-
ride, iron(III) chloride hexahydrate, copper sulfate pentahydrate,
barium chloride, acetonitrile (HPLC grade), N-methylsuccinimide
2−
2−
Determinations of SO
3
, iron ion, SO
4
ion, and the
cyclic voltammograms are carried out according to literature
8
methods.
(
NMS), succinimde, 2-pyrrolidinone.
Acknowledgements
Testing of the catalytic activity of the iron salts—organic solvents
We are grateful to US Filter for financial support and
Dr J. A. Crayston (University of St Andrews) for assistance with
the cyclic voltammetric analysis.
All redox reactions were carried out in a two-chambered glass
apparatus, 1-liter reactor provided by US Filter. The pH and redox
This journal is © The Royal Society of Chemistry 2006
Dalton Trans., 2006, 1147–1156 | 1155

View Article Online
References
8 P. Hemmes, L. D. Rich, D. V. Cole and E. M. Eyring, J. Phys. Chem.,
1
971, 75, 929.
1
G. X. Hua, D. McManus and J. D. Woollins, Comments Inorg. Chem.,
9 G. H. Khoe and R. G. Robins, J. Chem. Soc., Dalton Trans., 1988, 2015.
10 B. B. Hasinoff, Can. J. Chem., 1976, 54, 1820.
2
001, 22, 327.
2
3
J. Wieckowska, Catal. Today, 1995, 24, 405.
11 T. Sekine and T. Tetsuka, Bull. Chem. Soc. Jpn., 1972, 45, 1620.
12 R. D. Hancock and F. Marsicano, Inorg. Chem., 1980, 19, 2709.
13 C. Brandt and R. van Eldik, Chem. Rev., 1995, 95, 119.
14 M. Taqui Khan and G. Ramachandraiah, Inorg. Chem., 1982, 21, 2109.
15 M. M. Taqui Khan, Pure Appl. Chem., 1983, 55, 159.
16 M. M. Taqui Khan, A. Hussain, K. V. Subramanian and G.
Ramachandraiah, J. Mol. Catal., 1988, 44, 117.
17 M. M. Taqui Khan and R. S. Shukla, J. Mol. Catal., 1986, 37, 269.
18 M. M. Taqui Khan, R. S. Shukla and A. P. Rao, J. Mol. Catal., 1986,
39, 237.
F. C. Boston and M. L. Schneider, Proceedings, Gas Conditioning
Conference, University of Oklahoma, Norman, OK, April 1971.
B. Ferm, W. A. Freivald, D. McManus and M. Reicher, Sulfur Recovery
Using Non-Aqueous, Chelated Iron, Redox Catalyst Solution, in Sulfur
Recovery Conference, GRI, London, UK, 1997, vol. 8.
4
5
6
7
Q. Feng and H. Qaki, Polyhedron, 1988, 7, 291.
W. C. Bray and A. V. Hershey, J. Am. Chem. Soc., 1934, 56, 1889.
R. S. Sapieszko, R. C. Patel and E. Matilevic, J. Phys. Chem., 1977, 81,
1
061.
1
156 | Dalton Trans., 2006, 1147–1156
This journal is © The Royal Society of Chemistry 2006