Documente Academic
Documente Profesional
Documente Cultură
J. F. ALDER*
and K. KARGOSHA
Chemistry Department,
(Gt. Britain)
London,
SW7 2AY
SUMMARY
Conditions for the determination of total sulphur and disulphide/polysulphide
in petroleum products are described_ The sulphur is reduced by heating with sodium or Devardas
alloy under reflux with subsequent liberation of H,S and measurement of the chemiluminescent S, emission intensity in a hydrogen-argon
diffusion flame. The precision and
accuracy are good. Applications to light distillates and waxy residues are discussed.
232
imprecise method for the determination of sulphur The reasons for the
imprecision probably lay mainly in the loss of volatile sulphur species - the
recovery of sulphur from the samples was later estimated to be 50-60% and inadequate control of the heating and fusion procedures. The present
study was concerned in part with estimating the recovery of sulphur species
by this method, and with optimization of the pretreatment conditions in
order to maximize recovery and improve both precision and accuracy.
Disulphides and polysulphides play many very important roles in the wool,
synthetic polymer, rubber, petroleum and protein industries_ Most of these
roles are connected with the S-S bond and the presence of cross-linking sites.
The analytical difficulties which arise in the determination of disulphides and
polysulphides arise more from the type of samples encountered than from the
measurement techniques available. Many different quantitative methods have
been reported for the determination of disulphide and polysulphide in the
presence of other sulphur compounds [63 . These can be largely divided into
chromatographic (t.1.c. and g-c.) methods [7] and metal reduction followed
by chemical separation and determination
[8] _
The method of Wronski [9], which is probably the most recent chemical
separation-metal
reduction method, is based on conversion of disulphides
by sodium aluminium bis( 2-methoxyethoxy)dihydride
in benzene solution
to hydrogen sulphide and thiols which can be either directly titrated, after
dilution with alcohol, or separated by extraction and determined in the presence of each other by selective masking with acrylonitrile. This is a simple
and sensitive method: sulphur at the O-l-ppm level in hydrocarbons can be
determined with a relative standard deviation of 0.5-1.510, but the recovery
for di-t-butyl disulphide is only 75%. Metals such as zinc and sodium, and
hydrides such as sodium borohydride and lithium aluminium hydride, which
have been used for the reduction of disulphides and polysulphides to mercaptan
[lo],
suffer from poor reduction efficiency for tertiary disulphides, except for
the zinc reduction method under reflux conditions which is expected to yield
8%90% recoveF1 for di-t-butyldisulphide.
This recovery is, however, achieved
only by heating the sample for 1.5 h and purging for 3 h.
In the present study the metal reduction method worked out previously
was modified in order to differentiate between some of the various species of
sulphur found in petroleum products and to permit their determination by
chemiluminescent
flame photometry.
EXPERRUENTAL
Apparatus
and reagents
The hydrogen--argon
diffusion flame and burner assembly and generating
cell [5] are outlined in Fig. 1. The emitted radiation is passed through a
circular variable wavelength filter (Barr and Stroud, Glasgow) with a band
pass of about 20 nm at 348 nm. The filtered light is focused onto an E.M.I.
9781A photomultiplier
and the signal is passed via an amplifier and backing-off
233
r-l
H2/Ar/HZS
I I
Photomultplier
1 holnlng
Chimney
Senturn
Fd:er
**heel
I
I
Bypass
IF----Y
y!!
i-rgm
Ar
1
3.3
0.6
0.9
1.2
1.5
I.8
system.
Fig. 3. Optimization of flow rates. H, pressure, 1 p.s.i.; Ar pressure, 1.5 p.s.i. When one
curve was being obtained the other flow rates were at their optimum setttings.
234
The sodium sulphide solution was stabilized initially with a pH 11 borateNaOH buffer but a better method was to use aqueous 20% (w/v) NaOH. The
stock lOOO-ppm solutions thus prepared were stable for up to one week
(Table 1) but were standardized by iodimetric titration [ll]
before use.
Organic solutions were prepared fresh daily (Table 1). Refrigeration of the
organic solutions was unnecessary when solutions were prepared daily.
Improvement of the sodium fusion method for total sulphur determination
In the previous method for the estimation of total sulphur by sodium fusion
[5], the initial reduction of the sample was carried out in an open tube over
a flame. Although this permitted rapid analyses, results were not reproducible
and recoveries were poor. Attempts were therefore made to improve this opentube method and the subsequent liberation of H2S from the sodium sulphide
formed- Although recoveries of about 80% were obtainable, this did not give
good reproducible analyses. The poor recovery of sulphur in the reduction
process is due to the conflicting processes of sample vaporization on heating
the sample and of the reduction of the sulphur species which needs elevated
temperatures for reasonably rapid reaction.
To improve recovery, precison and duration of the reduction process, a
reflux method was devised_ The apparatus comprised a lo-ml three-neck flask
fitted with a nitrogen purge; the Liebig condenser was of single-surface type
(10 cm long) with a delivery tube which passed the purge gas into a lo-ml
round-bottom flask containing the absorbing solution_
A l-ml aliquot of a standard solution of di-t-butyl disulphide in liquid
paraffin or white spirit was placed in the reflux flask with a freshly chopped-up
2-mm cube of sodium. The flask was fitted to the condenser and the delivery
TABLE
Stabilityof standardsolutions
Aqueous
solutions
(initial
concentration,
100 ppm
S as Na,S)
In aqueous
solution
S found
(ppm) 2 s-d.=
In 20% (w/v)
NaOH solution
S found
(ppm) + s.d.
after 8 h
after 24 h
89 2 1
75 f 1
after 3 days
after 7 days
99.5
99.0
Organic
standard
solutions
In liquidparaffin
S added (ppm)
260
195
130
Talculated
for 4-6
(S added
as di-t-butyl
After 24 h
262 2 3
194 + 2
129 2 2
202 * 2
160 +- 1
107 2 2
analyses.
f 1
t 1
disulphide)
In white spirit
S added (ppm)
After 24 h
260
195
130
257 2 2
192 f 2
129 * 2
202 * 2
152 f 2
104 f 1
Optimization
Temperature
(C)
Liquid paraffin
380
345
272
Heating time
(min)
S recovery
W)
20
30
40
20
30
40
20
30
40
63
95
70
61
94
70
15
50
47
Temperature
(C)
S recovery
Heating time
(min)
(S)
20
30
40
20
30
40
20
30
40
67
89
73
66
90
74
30
60
40
White spirit
190
173
157
from liquid paraffm and white spirit, respectively. In both cases, therefore,
the lower temperatures were chosen for the analyses, so as to minimize loss
of volatile sulphurcontaining species. The reason for the decrease in yield
after about 30 min is the formation of an emulsion on mixing of the reduced
sample with aqueous sodium hydroxide and subsequent difficulty in separating
the aqueous layer.
Optimization of gas flows and stirring
The argon flows through the cell and into the flame, and the hydrogen
flow to the flame were optimized (Fig. 2). Hydrogen sulphide was generated
continuously by hydrolysis of a crystal of sodium sulphide with dilute hydrochloric acid. The results are illustrated in Fig. 3, which shows that the flow
rates are quite critical for maximum emission intensity from the flame.
The rate of stirring of the mixture in the gas generating cell was found to
be important. The stirring affects the rate of mixing of the analyte solution
and the hydrochloric acid in the vessel and the subsequent generation and
release of hydrogen sulphide. The recovery reaches a maximum just before
the onset of splashing in the cell and for optimum recovery the stirring speed
should be maintained just below this critical value.
Calibration curves
Calibration solutions were made up from di-t-butyl disulphide in white
spirit and liquid paraffin to represent low and high boiling petroleum products,
respectively, and from sodium sulphide in 20% NaOH. The solvent blanks
were 1.5 and 0.5 ppm S for the white spirit and liquid paraffin, respectively;
these were subtracted before plotting.
The calibration curves are shown in Fig. 4. The linear portion of the curves
lie in the range 5-70 ppm S in the original sample, corresponding to 0.5-7.0
r.cg
of sulphide in the loo-p1 aliquot taken. The shape of the curves is similar to
those obtained by other workers [12] ; the curves do not coincide because of
the different recoveries of sulphur. As the origin of the emission is chemiluminescent, it is not easy to say what causes the non-linearity. It may be due
to quenching effects in the higher concentration and kinetic effects in the
lower region where the S + S recombination may be incomplete; these points
have been discussed elsewhere 112,131 but warrant further investigation. The
recovery (and relative standard deviation) of the method compared to aqueous
solutions on 10 replicate analyses of samples containing 150 ppm sulphur was
95% (i7%) for liquid paraffin.solvent, and 90% (*S%) for white spirit solvent.
Analysis of petroleum products
Samples provided by B. P. Research Ltd. and previously analyzed by x-ray
fluorescence (x-r-f.) were analyzed by the reflux method. The samples had to
be diluted with white spirit for the light distillates and liquid paraffin for the
waxy residues, to bring the final solution into the correct concentration
range (Table 3). The results, obtained from calibration curves established
237
1.5-
3-
-5 -
2-
.5 -
I-
I
2
1.5
log cotxentrotlon
1.5
log concentrot8on
(ppm)
I
2
(ppm)
Fig. 4. Calibration curves for organic (A) and aqueous (B) solutions- Curve (a) is for liquid
paraffin solutions and curve (b) for white spirit.
TABLE
Analysis of petroleum
Sample
Light
8014
7935
8136
7914
7956
9143
9142
products
% S found (A s-d.)
9b S presenta
Sample
0.01
0.24
0.03
0.2
O-26
0.95
0.11
9246
9245
8630
8634
7990
7927
8635
distillates
Waxy
0.01 f 0.002
0.20 f 0.015
0.025 2 0.002
0.18 ? 0.01
0.2 +_0.025
1.05 + 0.06
0.14 + 0.01
X S presenta
residues
0.63 + 0.02
0.79 f 0.065
3.65 +_0.3
1.5 + 0.1
2.29 + 0.15
2.2 2 0.18
3.3 * 0.2
0.62
0.75
3.7 3
1.39
2.24
2.11
3.33
with organic solutions taken through the procedure, compare most favourably
with those obtained by x.r.f.
Determination
of dkulphide
and polysulphide
The reduction conditions with sodium are so drastic that all organic sulphurcontaining compounds appear to be converted to sulphides. Whereas total
Reducing agents
Magnesium_ Commercially
available magnesium
powder
was employed
Reductant
Signal intensity
(ax.)
Reductant
Mg
Zn
Zn
Zn
Unsuccessful
19
79
81
Zn amalgam
50% Zn-50%
SnClz
Cu (foil clippings)
Devardas alloy
(50% Cu, 45% Al, 5% Zn)
(fine powder)
(powder)
(granular)
(foil clippings)
aNormalized
to 100.
Signal intensity
(ax.)
81
10
23
loo=
Zinc. Powdered zinc gave low recoveries (Table 4) and was not studied
further. Granular zinc was much better and gave similar results to the zinc
amalgam and zinc foil. (The amalgam was made by grinding granular zinc in
liquid nitrogen, washing with water and treating with 2% (w/v) HgCl, solution;
it was stored under water.) The Zn--SnCl* mixture, which is a powerful
reducing agent, gave poor results.
Copper and Deuardas alloy. Copper clippings (<40 mesh) gave low results
but Devardas alloy (50% Cu, 45% Al, 5% Zn) was most efficient and was
therefore used in subsequent work_ Different weights of the alloy were
employed in the reduction of 1 ml of the sample solutions, and 100 mg proved
to be the optimum.
Temperature
and duration
of heating
The conditions of reduction particularly in the first part with the alloy
alone and then in the presence of hydrochloric acid are quite critical, In the
first step the reduction is thought to proceed via partial reduction of the S-S
bond and formation of metal sulphides. Extending the duration of the heating
at 320C during the first part of the reduction results in increased recovery
up to about 40 min (Fig. 5); to obtain this curve, the heating time during the
second stage was kept constant at 20 min.
The second stage of reduction which involves the alloy and hydrochloric
acid is thought to occur through two processes: decomposition of metal sulphides and also reduction of some S-S and S-H bonds to hydrogen sulphide
by hydrogen [lo] :
(t-Bu)-- -S-S-(t-Bu)
O.
0
10
20
U-U
--f 2(t-Bu)-SH
30
Heatmgttme
40
U-U
2(t-Bu)H
+ H,S
50
hnl
secondstage.
240
The optimum period of heating the solution during the second stage was
determined in the usual way and found to be 20 min at 120C. The amount
of concentrated hydrochloric acid employed in this step was varied in order to
obtain the optimum recovery of S. The concentration of the sodium hydroxide solution in the absorber was also varied in order to optimize the conditions;
this concentration should not be so high that the solution must be rendered
acid in the subsequent step, and the minimum concentration compatible with
good recovery is best. These experiments indicated the use of 2 ml of concentrated hydrochloric acid and 10 ml of 5% (w/v) sodium hydroxide in the
absorbing flask (Table 5).
During the heating process, the sodium hydroxide solution becomes warm
as the purge nitrogen passes through. This leads to loss of HIS (about 5-770)
and must be minimized. The flask was hherefore kept cool in an ice bath
thoughout the esperiment.
Optimization
of other parameters
When nitrogen was bubbled through the solution in the flask, lower recoveries (about 5%) were found, possibly because of more rapid removal of the
hydrogen from the solution. Passing the purge gas over the surface of the liquid
improved the recoveries; this use of purge gas is not always essential, but is
advisable in many cases to avoid atmospheric oxidation_
The order of addition of reagents in the reduction process was also investigated as well as the effect of omitting the Devardas alloy. The results obtained
(Table 6) show that the last method is by far the most efficient; this underlines the role that the reducing alloy plays in the initial reduction of the S-S
bond.
To check whether any H,S passed through the absorbing flask a second
flask was put in series containing the same 5% sodium hydroxide solution;
no sulphide was detected in this flask.
TABLE
Amount
HCI (ml)
0.5
1-O
1.5
2.0
2-5
of
of sodium hydroxide
in absorbing solution
Intensity (a-u.)
30
absorber
Aq. 2% NaOH
absorber
Aq. 5% NaOH
absorber
24
29
16
6
8
24
56
86
95
85
24
24
87
100
lOOa
241
TABLE
Method
Add
Add
Add
Add
2 ml 11
100 mg
2 ml 11
100 mg
<2=
8
19
1OOb
Interference
studies
242
TABLE
Intensity
(au_)
Compound
Liquid paraffin
Di-t-butyl disulphide
Ethylmercaptan
Diphenyl trisulphidea
Diphenyl sulphide
<l
100
110
95
<1
Dimethyl sulphide
Dimethyl sulphoxide
Benzene potassium sulphonatea
1 ,3-Dithianea
aDissolved fist
TABLE
Intensity
(a.u.)
2
0.03
2
1
Intensity
(a-u.)
Interferent
1ooa
>350
110
130
145
[Di-t-butyl disulphide]
Ethyl mercaptan
Ethyl mercaptanb
Diphenyl sulphide
Dimethyl sulphide
Intensity
(au.)
Interferent
Dimethyl sulphoxide
Benzene potassium sulphonate
1,3-Dithiane
Elemental sulphurC
Elemental sulphurc* d
96
96
100
> 350
98
- loo
b--e--:
go-.A-.-&-.-&
3 B
s2=)-
70
Knxl
2000
Sulphde-S
3ooo
4000
(ppd
on disulphide determination.
Conclusions
This work has shown that the combination of metal reduction of sulphur
compounds to sulphides in petroleum products coupled with release of
hydrogen sulphide into a cool flame and measurement of the S, emission
provides a useful method for the determination of total sulphur and for the
separate determination of sulphur species. The method is capable of good
243
TABLE
S in
diluted
sample
(ppm)
Di-t-butyl
Added
(ppm)
Recovery
(X)
Added
(ppm)
Recovery
(a)
7866
7914
7956
7957
8110
9142
9143
400
800
520
330
450
300
950
320
420
250
200
180
150
120
90
92
91
93
90
90
91
320
410
240
200
180
150
120
89
90
92
92
91
91
90
disulphide
Diphenyl trisulphide