MICROCHEMICAL

JOURNAL

41, 139-147 (1%)

Automatic Absolute
Combustion

Determination
of Total Sulfur
and Coulometric
Titration

by

KEVIN D. WILLS,'EDWARD W.D. HUFFMAN,~R.,AND

EDWARDW.D. HUFFMAN, JR.
Huffman

Laboratories,

4630 Indiana

Street, Golden, Colorado

Received October 11, 1989; accepted November

80403

3, 1989

An automatic method for the absolute determination of total sulfur, which overcomes the
tendency of previous methods to give low sulfur recoveries on samples with high oxygen
demand, is described. The technique combusts the sample in oxygen and sweeps the combustion gases through a hydrogen peroxide trap to remove the sulfur oxides from the gas
stream. The trap contents are volatilized and swept with nitrogen through reduced copper
heated to 890C. Sulfur oxides are quantitatively
converted to sulfur dioxide and titrated
coulometrically
with iodine. Halogens are held by the copper and do not interfere with the
sulfur determination. Analysis time is 12 min and automation of the method allows unattended operation. Sample size is 0.5 to 20 mg. Sulfur levels of 0.05 to 100% can be determined. 0 1990 Academic Press, Inc.

The determination of total sulfur by combustion is complicated by the fact that

sulfur is not readily oxidized to a single species. When combusted with oxygen in
an empty tube, sulfur is oxidized to a mixture of SO, and SOs. Even under
conditions which are theoretically ideal for the production of SO,, the SO,:SO,
ratio can vary from 70 to 90%. Consequently, methods for determining sulfur by
detecting the SO2 produced by open tube combustion systems, must be empirically calibrated and the calibration factor may change with varying sample type.
Interferences from halogens are an additional difficulty associated with the open
tube combustion sulfur determination.
Dugan (I), Kirsten (2), Kirsten and Nordenmark (3), and Paroutaud er al. (4)
successfully use copper at 800-900C to remove interfering halogens and to quantitatively reduce SO, to SO,. However, in these methods the amount of oxygen
used for sample combustion must be limited or the Cu supply will rapidly be
oxidized. Because of this, coals and other samples requiring large amounts of
oxygen for combustion frequently gave low sulfur results in our laboratory. In
addition, pyrolysis gases, formed at the moment of sample introduction, may
reduce metals in the combustion zone if sufficient oxygen is not available. The
presence of these reduced metals may permit the formation of stable metal sulfides and thereby cause a low sulfur value (2, 3). These problems can, in theory,
be overcome by using smaller sample weights, but this is not an acceptable solution if the sulfur levels in the sample are very low, or if the samples are not
completely homogeneous.
To whom correspondence

should be addressed.
139
0026-265x/90 $1.50
Copyright
0 1990 by Academic Press, Inc.
All rights of nproduction in ally form rwtxved.

140

WILLS,

HUFFMAN,

AND

HUFFMAN

Combusting the sample in an unlimited supply of oxygen eliminates these problems without limiting the sample size. We have developed a method with the
desirable features of unlimited oxygen for sample combustion, quantitative conversion of all sulfur to Sot, absolute measurement of SO2 by coulometry, and
unattended automated analysis of preweighed samples. This modification of the
previous methods involves combusting the sample in oxygen at 1080C and trapping the SO2 and SO, from the combustion gases. Oxygen is then purged from the
system with nitrogen. The oxides of sulfur are released from the trap and quantitatively reduced to SO* by elemental copper at 890C. Finally, the SO2 is titrated
coulometrically with iodine. Halogens, which are potential interferences with the
coulometric titration, are held by the copper.
Quartz wool moistened with a 6% hydrogen peroxide solution makes an excellent trap for SOZ and S03. In the trap, SO* is oxidized to SO3 by H,O, and SO,
is held as H,S04. Upon heating, all HzO, H,Oz, and H,S04 is volatilized and
swept through copper. Water vapor passes directly through the copper and HzOz
is reduced to H20. Sulfuric acid is decomposed to Hz0 and SOS. Elemental
copper reduces SO, to SOZ.
EXPERIMENTAL
Apparatus
Valves and tubing. Three-way electrically actuated Teflon valves (General
Valve) are used. Teflon tubing is used throughout the instrument. Stainless steel
and other materials have a tendency to react with and retain SO*.
Reaction tubes. Two tubes made of quartz are required. Both tubes are illustrated in Fig. 1. The inner tube fits coaxially inside the top of the outer tube and
catches the sample as it falls into the combustion zone. It is designed to be easily
replaced without disturbing the rest of the system. In this way sample residues can
be removed after a day of runs and the more complex outer tube is spared attack
by these residues. A layer of crushed quartz chips placed in the inner tube aids in
mixing the combustion gases with oxygen to ensure complete combustion. The
lower section of the outer tube is filled with reduced copper and the trap side arm
is packed with tine quartz wool. The trap can be packed through the 6-mm tube,
but it is much easier to pack the trap before the glassblower attaches it to the tube.
The quartz wool is an absolute necessity and must be packed tightly enough to
trap any sulfuric acid mist that is formed (6). A 25mm Ultra-Torr (Cajon) in-line
fitting is used to connect the two tubes together.
Furnaces. Two furnaces are required. The combustion furnace must be capable
of operating at 1080C and the reduction furnace at 890C. This apparatus uses
30-cm-long tube elements (Watlow) insulated with ceramic fiber and mounted
vertically. The furnaces must be positioned apart such that the combustion tube
trap arm may be routed out to the side. However, the furnaces must not be so far
apart as to create a cool zone capable of condensing sulfuric acid. Ceramic fiber
material is used to till the gap between the furnaces after the tube is in place. The
trap arm must extend far enough out from the furnaces to allow cooling water to
be sprinkled on the trap between runs.
Pumps. A precision pump (Fluid Metering) constructed of ceramic is used to

DETERMINATION

141

OF TOTAL SULFUR

INSIDE
FLOW

6 CM

OUTER
TUBE
A-

3 CM
I4CM

INNER
TUBE

5CM

4CM

43CM

4 CM

16 CM

FIG. 1. Quartz reaction tubes. The inner tube fits into the top of the outer tube. (A) 2%mm-o.d.
quartz. (B) 15mm-o.d. quartz. (C) 6-mm- o.d. quartz. (D) Ill-mm-o.d. quartz. (E) II-mm-O.D. quartz.
(F) Quartz wool. (G) Crushed quartz chips. (H) Metallic copper.

pump a fixed volume of 6% hydrogen peroxide to the trap. An inexpensive recirculating pump is used to sprinkle water on the trap to cool it rapidly between runs.
Trap heater. Seven meters of 25gauge Chromel-A wire are wound directly
along the entire length of the trap arm. The heater power is adjusted with a variac
connected to 120 V ac.
Water collector. The water from the trap must be removed from the gas stream

142

WILLS, HUFFMAN,

AND HUFFMAN

before entering the coulometer cell. This is accomplished with the borosilicate
glass collector shown in Fig. 2. A small electric oven keeps the collector at 70C.
Elevating the temperature lowers the solubility of SO* in water sufftciently to
prevent the retention of SOz by the water in the collector.
Flow meters. Five flow meters (Porter) with needle valves and scales of 200
ml/min air are used.
Autosampler. A lOO-slot autosampler was constructed at Huffman Laboratories. It consists of a carousel to hold the samples, a plug valve to admit the sample
to the combustion zone, and motors to drive the carousel and valve. Connection
is made to the top of the inner reaction tube with a 25-mm Ultra-Torr in-line
fitting.
Computer. A Vitrax IV microcontroller (Sintec) is used in conjunction with
solid state relays (Opto-22) to control the valves, autosampler, trap heater,
pumps, and coulometric titrator. The computer is programmed in Basic and the
program is permanently stored in EPROM.
Coulometer. An unmodified sulfur coulometer (UIC) and its reagents are used.
The coulometer is described by Greyson and Zeller (5).
Reagents
Copper. Cupric oxide (EM Science) is reduced at a temperature of 400C with

hydrogen. It is very important for the copper to have a minimum of metallic

impurities (3) as they will retain sulfur, producing low results and memory effects.
Hydrogen peroxide. It is equally crucial that the H,Oz used contain no impurities that may form sulfur compounds which are difficult to decompose. The 6%
solution required is made fresh daily from 30% hydrogen peroxide (SargentWelch).
A
2 CM
1
Ir

7CM

2 CM
t
FIG. 2. Water collector. The collector is made of borosilicate glass and is kept at a temperature of
70C. (A) 6 mm o.d. (B) 20 mm o.d.

DETERMINATION

OF TOTAL

SULFUR

143

Vanadium pentoxide. To assist in the decomposition of inorganic sulfur compounds a generous quanity, 10 mg, of V,Os (Alpha Resources) is added on top of
the sample in the sample container.
Tin capsules. Samples are contained in tin capsules (Ludi) 3.5 mm in diameter
and 9 mm in length.
Procedure
The nitrogen supply is set with a regulator to 5 kPa. The inside and outside
combustion tube flow rates are set to 40 mUmin with the flow meters labeled Fl
and F2 in Fig. 3. Valve Vl is switched to the oxygen supply and the oxygen
pressure is adjusted to roughly twice that of the nitrogen in order to produce a
flow rate of 80 ml/min without disturbing the setting of Fl or F2. Switch the valve
back to the nitrogen position before heating the furnaces.
A glass bubbler is filled with 10 ml of methanol and inserted in series between
F4 and V3. Methanol vapor serves to keep the copper reduced between runs. This
greatly extends the life of the tube by renewing the supply of metallic copper and
by eliminating copper oxides which can attack the quartz tube. Methanol is a fire
hazard around the high temperature furnaces, so use only small quanities in the
bubbler. Adjust F4 for a flow of 40 ml/min up through the copper and out the vent.
Also, adjust flow meter F5 to 10 mYmin into the water collector and up into the
copper.
With the vent open, sampler valve closed, nitrogen flowing through Fl and F2,
and nitrogen/methanol flowing up through the copper, bring the combustion furnace temperature up to 1080C and the reduction furnace up to 890C. Higher
reduction furnace temperatures cause excessive sintering of the copper.
A determination is begun by switching Vl to oxygen and filling the combustion
zone with oxygen for 1 min. At this same time, the vent is open and nitrogen/
methanol is flowing up through the copper. In this way the copper is not oxidized.
Also, the trap heater is off and is being sprinkled with water from a recirculating
pump to cool it from the previous run. A small catch basin directly underneath the
trap returns water back to the pump reservoir.
After the combustion zone has been tilled with oxygen, the sample valve is
opened but the carousel is not advanced so that the sample does not fall. Valve V2
is switched to admit nitrogen from F3 which has been adjusted for a flow of 150
ml/min. The high flow serves to blow the peroxide solution into the trap and
prevents backflow should the trap contents bump when they are volatilized. The
peroxide pump is switched on for 5 s and pumps 1 ml of hydrogen peroxide
solution into the trap. Valve V2 is left on for an additional 2 s and then switched
back to the vent position. The carousel is advanced, dropping a sample into the
combustion zone, and the sampler valve is then closed.
Oxygen continues to flow through Vl and nitrogen/methanol through V3. The
sample is combusted in this state for 2 min. All SOZ and SO3 produced during
combustion is now being trapped as combustion gases are swept out the vent.
Valve Vl is then switched to nitrogen for 1 min. This purges oxygen from the
combustion zone in preparation for the reduction phase.
To initiate the desorbtion of sulfur from the trap, valve V3 is switched to allow

WILLS, HUFFMAN,

AND HUFFMAN

AUTO SAMPLER

VENT
F3

ONE
WAY

t
L-l-7

..-

.01

REDUCTION
89OC

I-----I;

Hj--

F5

COULOMETER
+

DRAIN
v4

FIG. 3. Instrument diagram. Fl through FS are 150 cc/min flow meters. Vl through V4 are threeway Teflon solenoid valves. Not shown are a small recirculating pump and a catch basin beneath the
trap side arm for cooling the trap with water.

flow to the coulometer and valve V2 to admit nitrogen into the trap. The cooling
water is shut off and power is applied to the trap heater. The heaters power is
adjusted with a variac to produce a bright red illumination. However, during the
first minute of heating the heater is not operated at this level. It is operated at half
power by turning the heater on for 1 s then off for 1 s in a cyclic manner. This is
to prevent bumping as the trap contents are volatilized. The trap contents are
swept down through the copper by nitrogen flows from Fl, F2, and F3. The
oxides of sulfur are quantitatively reduced to SOZ.

DETERMINATION

OF TOTAL

SULFUR

145

Water vapor from the trap solution rapidly condenses as small droplets in the
Teflon lines upon leaving the reduction furnace. These droplets are collected by
the water collector and are prevented from entering the coulometer cell. The small
nitrogen flow through FS serves to keep water droplets out of the collector drain
line where they cannot be easily swept free of dissolved SOZ.
The coulometer, which has previously been reset, should begin titrating SOZ.
After 1 min of half heating power the trap is left on continuously to apply full
power to the trap for 7 min. At the end of this period the coulometer is read and
valve V4 is switched to allow water in the water collector to be blown out the
collector drain for two seconds. The cycle is then repeated for the next sample.
RESULTS AND DISCUSSION
Results obtained in the sulfur determination of a variety of materials of known
sulfur content are summarized in Table 1. The immunity of the system to interferences by halogens is illustrated in Table 2. Of note is the excellent sulfur
recovery from a 20-mg sample of coal reference material 1632b listed in Table I.
We have found such a recovery, for so large a sample weight, to be impossible to
achieve with the previous techniques because of the limited availability of oxygen.
To quantify the accuracy and precision of the technique, the mean percentage
of sulfur determined for 39 thiourea (42.11% S) samples was calculated to be
41.57% S with a standard deviation of 0.44% S. The percentage relative accuracy
of these 39 determinations is therefore 1.3% with a percentage relative standard
deviation of 1.1%. Some of the observed deviation appears to originate in the
coulometer. The coulometer uses an amperometric method of indicating the titration endpoint and poor precision will result if the iodine generation current
affects the endpoint indicating current. The positioning of the sensing and generating electrodes is critical in avoiding interactions between these two currents.
The generating electrodes must be as close as possible to the frit separating the
anode and cathode compartments. However, if the generating electrodes are too
close to the frit, the generated iodine will not be immediately stirred into the bulk
solution and the endpoint will tend to be overshot. A compromise position can be
found which is workable although it does not always provide satisfactory precision.
There is an apparent negative bias of 1 to 3% relative in the sulfur recoveries
listed in Table 1. It is difficult to assessthese data against the results generated by
other methods (Z-3) as they use empirically calibrated detectors which can hide
the bias with the calibration factor. Paroutaud et al. (4), however, uses an absolute
coulometric titration that does not appear to have the negative bias that is evident
in our data. The calibration of our coulometer was checked by placing a silverplatinum cell in series with the sulfur titration cell. By measuring the weight loss
of the silver electrode, it was verified that the electrical calibration of the sulfur
coulometer has a percentage relative accuracy of 0.1%. Therefore, the calibration
of the coulometer is not the source of the bias. Inefficiencies in the titration cell
should cause high results and so cannot be the source either. Additional investigation of the method is necessary in order to identify the origin of this negative
error.

146

WILLS, HUFFMAN,

AND HUFFMAN

TABLE 1
Sulfur Recoveries
Sample type
(Grade)
Thiourea (REAG)
Thiourea
Thiourea
DBDS (PSTD)
DBDS
DBDS
BaSO, (REAG)
BaSO,
BaSO,
Na$iO, (REAG)
NGQ
NGQ
Sulfur (UNKNWN)
Sulfur
Sulfur
Sulfur
sulfur
SRM2683
SRM2683
SRM2683
SRM1632b
SRM1632b
SRM1632b
Sulfanilamide (PSTD)
Sulfanilamide
Sulfanilamide
Phenylthiourea (PSTD)
Phenylthiourea
Phenylthiourea
Diphenylsulfone
@EAG)
Diphenylsulfone
Diphenylsulfone
BITUHCI (UNKNWN)
BITUHCl
BITUHCl
Note.

PSTD,

primary

from Various Materials

Weight
(CL&

Theory
@ 9

Found
@ S)

% S recovered
(% found/% theory)

4418

42.11

41.78
41.81
41.87
25.62
25.81
26.23
13.57
13.50
13.50
25.54
25.76
25.50
93.78
94.95
94.88
94.66
95.19
1.90
1.83
1.87
1.91
1.89
1.82
18.32
18.36
18.32
21.15
21.41
20.92
14.44
14.35
14.38
15.37
15.42
15.29

99.21
99.29
99.43
98.45
99.18
100.82
98.77
98.27
98.26
100.40
101.24
100.22
93.78
94.95
94.88
94.66
95.19
102.70
99.16
101.08
101.06
99.92
96.30
98.39
98.60
98.40
100.37
101.61
99.28
98.33
97.69
97.67
97.14
97.49
96.65

3774
6651
4704
2351
7167
585.5
5066
4631
4673
4570
709
2129
960
3437
4734
11286
15662
14407
13134
20038
9582
5259
3913
5520
5414
3611
4536
3857
4498
3652
3184
3552
4249

26.02

13.74

25.44

100.00

1.85

1.89

18.62

21.07

14.69

15.82

standard grade; REAG, reagent grade; UNKNWN,

BITUHCl,
benzylisothioureahydrochloride.
a SRM2683 and SRM1632b are NBS-certified
coal standards.

grade is unknown;

DBDS, dibenzyldisulfide;

Vanadium pentoxide is used liberally on all samples. Kirsten (2) states that
sulfur may be retained by the tin capsules and by metallic impurities in the tin and
uses WOs to eliminate this problem. We experience difficulty with the W03 itself
retaining sulfur, a possibility mentioned by Kirsten et al. (3), and have instead
been successful using V205. When analyzing highly stable inorganic sulfur compounds such as barium sulfate it is helpful to place a layer of V205 on the bottom
of the capsule before adding the sample and then place an additional layer of V,05
on top of the sample.

DETERMINATION

147

OF TOTAL SULFUR

TABLE 2
Sulfur Recoveries of Halogen-Spiked Thiourea Samples
Weight (pg) of
thiourea (42.11% S)

Weight (pg) of
HCB spike

% Sulfur recovered
(% found/42.11%)

3815
4133

5684
4848

100.45
99.59

a Hexachlorobenzene.

Except for some perceived precision problems, the coulometric sulfur titrator is
very convenient to use. Empirical calibration constants normally associated with
volumetric titrations and other measuring techniques such as thermal conductivity
have been eliminated as the sulfur coulometers calibration is based on the absolute standards of the ampere and the second. Interfacing the coulometer to a
computer and installing an autosampler on the instrument has created an instrument with a great ease of operation. Although the determination time of this
technique is 12 min, the operator can complete a large number of determinations
by letting the instrument run unattended overnight.
ACKNOWLEDGMENT
The sulfur coulometer used in this study was kindly provided by UIC Inc., Joliet, Illinois 60434
0863.

REFERENCES
1.
2.
3.
4.
5.
6.

Dugan, G. Anal. Let?., 1977, 10, 639-657.

Kirsten, W. J. Anal. Chem., 1979, 51, 1173-1179.
Kirsten, W. J.; Nordenmark, B. S. Anal. Chim. Acta, 1987, 196, 5!3-68.
Paroutaud, P.; Cousin, B.; Fraisse, D. Microchem. J., 1980,25, 267-280.
Greyson, J.; Zeller, S. Amer. Lab., 1987, 9(7), 44-51.
Hallet, L. T.; Kuipers, J. W. Znd. Eng. Chem., Anal. Ed., 1940, 12, 357-359.