The evaluation of the arsenator
Dr. Walter Kosmos
Department of Analytical Chemistry, Graz University,
The idea to develop a field determination for arsenic goes back to my
first visit at West Bengal in 1995. I realized that a simple instrument
is needed to determine the arsenic concentration in ground water right
at the source. One of my students accepted this problem for his master
thesis. Unfortunately the principal idea to keep the system as early as
possible was lost during time. The reason was that by adding improvements,
better results were achieved. As analytical chemists always look for the
best results, and the method at the end was not a simple field kit but
a field instrument. Although the design with all the electronics became
more and more complicated, the handling was kept as simple as possible
to allow the use for less skilled people.
At the moment when I was aware that such an instrument is needed in
large number I organize a group for the development of commercial design,
including specialists for programming, electronics and mechanics. As it
was not possible to build a system with a latest design ready to take it
to Bangladesh in September, the model used there was a prototype. This
prototype has only a different casing, all the function and performance
is the same as for the final instrument. On the following graphic the two
main part can be observed: the main body with the electronics and stirrer,
and on top of the Erlenmeyer flask which contains the sample, the actual
The general measurement is based on the generation of arsenic, AsH3,
and its subsequent detection via the reaction with mercuric bromide, which
is known since decades as the Gutzeit method. Arsenic reacts with mercuric
bromide to coloured products according the following equation:
The structure of the reaction product is not exactly known.
The product has a light yellowish colour, further reaction with arsenic
leads to brownish products. For the original Gutzeit method, also known
as mercuric-bromide stain method, the gas stream containing the arsenic
passed an impregnated paper strip placed in small glass tube on top of
the reaction flask. At the end of the reaction the length of coloured zone
should be proportional to the quantity of arsenic in the reduction flask.
Many parameters influence the result and not surprisingly, accepted results
can be excepted only under painstakingly controlled conditions. With any
analytical method, good accuracy and precision at low doses are generally
achievable only when the detection limit of the method is not higher than
one-tenth of the lowest concentration of interest. Intensive experiments
at our institute showed that that stain method is unsuitable for the quantification
of arsenic at concentrations below the Bangladesh standard of 50 parts
per billion (ppb), even if a large error (>65%) is accepted. Therefore
this method can certainly not be used to quantify arsenic at concentrations
< 100 ppb.
To overcome these problems we have taken into account
in our method following basic configuration:
The arsenic gas should not pass by the impregnated strip
rather pass through it. This assures the reaction of all the arsine gas
with mercuric bromide and enhances before the detection limit. The poor
estimation of the coloured strips by human eye is replaced by a light absorption
measurement, achieved by a diode emitting light in the proper wavelength
range and a photodiode arranged in transmission mode. Both objectives were
realized in the measuring part on top of the flask.
The Arsenator switched on and a message start analysis
is readable on the display. After pushing the OK buttons the following
message appears add sample select volume. Depending on the
expected concentration a sample volume of 1, 5 or 50ml could be selected.
Directly at the tube well the water is filled into a calibrated cylinder.
At expected concentration from zero to 50 ppb a sample volume of 50 ml
is transferred into the Erlenmeyer flask. Pushing OK the message insert
filter appears. The measuring part is opened and the impregnated filter
paper is inserted using tweezers and the cell is closed. Pushing Ok instrument
starts to calibrate itself to zero. Pushing OK button again the program
asks to add the reagents. First 10 ml of concentrated hydrochloric acid
is added, and second a small quantity of NaSCN diluted in NaCl. Pushing
OK the stirrer starts automatically and runs for five minutes. The reason
for this step is to reduce all arsenate to arsenite, because arsine gas
can only evolve during reduction by metallic zinc from arsenite. After
the five minutes, the stirrer stops and the display asks to insert cotton
and after wards to add the reductant. The cotton pad is impregnated with
lead acetate and filters the sulphur hydride from the gas steam which interfere
the detection. Immediately after addition of the reductant (metallic zinc)
the measuring part is put on the top of the flask and after the OK button
is pressed the stirrer starts again for twenty minutes. After the reaction
time the instrument measures the light absorption of the filter depending
on the amount of the yellow reaction product. The electronics process the
signals and with the use of a stored calibration curve the result is display
Test for Precision and Accuracy with Real Water Samples at Bangladesh
The evaluation was done with water samples taken from
tube well located at the police training center near Khulna. The analysis
was performed on 7th September, 1998. The following results
were obtained (all in ppb)
|Arsenator in field
|DPHE–Ag DDC prev. meas.
|DPHE–Ag DDC 9.9.98
The results clearly show the low determination limit for
real samples, e.g. STW-4, which is programmed for this device with 0.5
ppb, carried out with a sample volume of 50ml. The actual detection limit
is 0.2 ppb, but for the practical reasons limited to 0.5. Lowering the
sample volume to 5 or 1 ml allows a wide analytical range, e.g. STW – 3
and STW- 5 experiments in the laboratory showed that arsenic concentrations
can go up to 2000 ppb for a ml sample.
The results agree well with the laboratory results of
the DPHE lab at Khulna using the Ag-DDC standard method.
As we can see from the graph the results for the stock
solution differ by 17%. As no third independent method was available it
was not possible to determine the actual concentration of the stock solution.
There fore all results differ relative to this discrepancy.
For further testing the larger quantities of water from
the tube wells ST-4 and STW-5 were taken to the DPHE laboratory.
Measurements of Repetitions
The water sample STW-5 was measured five times in the
DPHE lab with the Arsenator. It should be noted that it is well known that
water samples always show lower values after transportation. Quite a lot
of silicate and oxides precipitate during transportation and storage. It
may happen that they absorb amounts of arsenic even if the sample is acidified
to prevent the iron from precipitation as iron hydroxide.
The results are following:
These six repetitions with the same sample lead to a mean
value of 231 ppb and a standard deviation of 8 ppb, which corresponds to
an analytical variability of about 3% relative. The precision is also demonstrated
by the fact, that the sample volume plays not a great role for the result.
At this point I may remark that the precision of the Arsenator
is equal or even better to that of much more sophisticated methods like
hydride generation atomic absorption spectroscopy (HG-AAS).
Precision by Standard Addition
With the low water arsenic sample taken from STW- 4 I
performed standard addition measurement at the DPHE laboratory. To the
50-ml sample various amounts of dilute stock solution was added. As mentioned
above the accurate concentration was not known but this fact as a constant
has no influence for linearity.
|Stock added (ul)
As shown the linearity is excellent and all results are
within the 95% confidence interval.
As the results with real samples have demonstrated, the
Arsenator method is reliable, accurate and precise method although it is
a field method. The performance found in Bangladesh was the same as shown
at other demonstrations like at the US Geological Survey at Boulder where
samples with heavy salt load and acid mine effluents showed the same precision.