Dissertation

Department of Chemistry
Final Year Project 2015-2016
(Module CH2325)
Development of detectors for pseudo-
chemical warfare agent formazan
complexes
Conor Jay Rees

Supervisor: Dr Joseph Beames
Contents
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.iv
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..1
1.1 Introduction to the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Formazans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Structures of Formazans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Tetrazolium salts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..8
1.5 Synthesis of Tetrazolium salts and Formazans . . . . . . . . . . . . . . 10
2. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Experiment 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
3.1 Red 610nm long pass filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Orange 550nm long pass filter . . . . . . . . . . . . . . . . . . . . . . . . . . .16
3.3 Green 495nm long pass filter . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
3.4 Blue filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
3.5 Red 610nm long pass filter (precise measurements) . . . . . . . . . .21
3.6 Orange 550nm long pass filter (precise measurements) . . . . . . . 22
3.7 Green 495nm long pass filter (precise measurements) . . . . . . . . 23
3.8 Blue filter (precise measurements) . . . . . . . . . . . . . . . . . . . . . . . 24
3.9 Assignment of wavelengths to micrometer positions . . . . . . . . . 25
3.91 Analysis of Formazan complexes . . . . . . . . . . . . . . . . . . . . . . . 26
3.92 Boxed Analysis of Formazan complexes . . . . . . . . . . . . . . . . . 28
3.93 Analysis of colour change using coded image analysis . . . . . . 32
4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
ii

Acknowledgements
I would like to thank Dr. Joseph Beames for his continuous support advice, expertise
and allowing for me to use your lab. Thank you for allowing me to be apart of this
project.
iii

Abstract
The development of a detector in which could detect a pseudo warfare formazan
complex was approached by first a compact built spectrometer. Wavelengths were
assigned to micrometer positions on the spectrometer in order for analysis of pseudo
warfare formazan complexes at different dilutions. The sensitivity of the spectrometer
was tested against the different dilutions and was proven to stop giving accurate
readings at a dilution factor of 4.
The second approach was a coded program which assessed the RGB value of an image,
while then using ‘kmeans’ clustering and thresholding to alert the user if one was
exposed to toxins and to inject an antidote.
1. Introduction
iv

1.1 Introduction to project
The aims of this analytical chemistry project were to begin developing a sensor to detect a formazan
complex, which had been formed by reducing a tetrazolium salt using a pseudo chemical warfare agent,
which gives an intense red colour when a warfare agent is present (reaction scheme shown in Figure 1).
The final reaction between the F2Tz tetrazolium and the analyte does not have a definitive mechanism,
however a photogeneration has been reported (shown in Figure 2), this proceeds via two intermediates
within seconds.i
In this sequence conducted by F. Kanal et al. upon the 2,3,5-triphenyl-tetrazolium chloride (TTC), in
water the 2,3,5-triphenyl-tetrazolium formazan (TPF) and the polycylic 2-phenyl-benzo[c]tetrazole[2,3-
α]cinnolin-4-ium (PTC) is formed under UV irradiation. However when the solvent is an alcohol only
PTC is formed.
Fe
B
O
O
Ph
Ph
B
O
O
Ph
Ph
Fe
B
O
O
Ph
Ph
B
O
O
Ph
Ph
F
F
2-
Fe
B
O
O
Ph
Ph
B
O
O
Ph
Ph
F
F
1-
Me
P
O
F
O
4
2
2 2
4F-
base nucleophile
N
N N
N
F
F Cl
Colourless F2Tz Tetrazolium
N
NH N
N
F
F
Intense Red Formazan
A
Figure 1: Example of Detection Chemistry of a G-agent systemii
v

Figure 2: Photochemical reaction pathway of TTC, products of TPF and PTC, also the intermediates ro-TTC, ro-
TTC’ and TTC’1
The TCC intermediates in Figure 2 have different numbers of π electrons as the TCC starting has 6,
intermediate 1 has 6, intermediate 2 has 7 and so does intermediate 3, while TPF has 6.
In order to determine the presence of these warfare agents, a HGV (Highly selective, Greatly enhanced
responses & times, Very easy to use) swab test was invented by Ian Fallis, 20152
(shown in Figure 3).
The swabs have been patented and are able to detect the G-agent through the colour change chemistry
discussed above.
Figure 3: HGV swab test kit
However the shortcomings with these swabs is they do not indicate if or when the individual would
need to inject an antidote to combat the poison they have been exposed too. Due to the severity of these
vi

warfare agents the user would not want to inject said antidotes unless it is vital, as they themselves can
be detrimental to ones health. Furthermore it would be impractical to continuously swab surfaces and in
hope of an outcome with no designed timeframe. Finally the swabs rely on the evaluation of the human
eye. The problem there being the reliance on human judgment to determine this colour change could
prove to be problematic, human error as it is. Because of this difficulty to determine slight variances in
colour over a sustained period of time, that are vital in indicating small amounts of warfare agents that
may be present. Conducting an experiment such as this in a high stress environment (e.g. warzone,
conflict, laboratory etc.) can prove extremely challenging. Thus more advanced methods are needed in
order to carry out said experiments to a high standard whilst maintaining stringent safety procedures.
Two solutions to this problem were approached in this project; the first approach would be to write a
program, which utilizes the inherent colour sensitivity of a CMOS/CDD camera (most smart phone
cameras are suitable) to detect a colour change from the starting solution to the final, which contains
the warfare agent. This program would recognize the colour change and alert the user to the danger in
real time, while using the smart phone camera would reduce the equipment load.
The second method approached was to provide a spectroscopic wavelength separation before the CMOS
detection, where is assumed that the intensely coloured formazan would absorb at certain wavelengths
and would be able to be identified. Again this method would use a mobile phone as a CMOS sensor.
This would possibly be more accurate, however it would require much more equipment than the first
method and may be detrimental to the speed of analysis.
1.2 Formazans
Formazans are made from the reaction of tetrazolium salts by dehydrogenases and also reductases
(illustration shown in Figure 4). These are compounds have a usual characteristic chain of:
N=N-C=N-NH:iii
Figure 4: Reduction of Tetrazolium salt to formazan complex
The history of these tetrazolium salts and formazans being synthesized can be found to be documented
over a century ago by Friese in 1875. Friese had made his formazan by reacting benzene diazonium
vii

nitrate with nitromethane, this produced a product called “Neue Verbindung”, a cherry red product, this
was known as the first formazan.iv
9 years later in 1894 von Pechmann and Runge had described the
synthesis of particular tetrazolium salts by oxidating formazan compounds.v
From methods described by
Pechmann and Runge hundreds of tetrazolium salts were prepared in the years that followed their work.
Formazans have been documented in literature since the rise of said work and were continuously
studied throughout the 1900s, however due to lack of modern techniques in order to characterize them
there was little known about their properties apart from their intense colour, this was how they were
identified in the 1900s. These colours usually range from red to orange or blue depending on their
different structures.
Formazan compounds have a certain class and characteristics and were named “formazyl compounds”
by Bamberger and von Pechmann. The term guanazyl was later applied to variations of the formazan
structure, as the =N-NH- group is attached to a guanyl group (in formazan compounds this group is
attached to an aryl radical).
There the guanazyl compounds will look like (shown in Figure 5):
-N=N-C=N-NH-C(=NH)-NH2
Figure 5: Guanazyl Compound
However the chemistry of these types of compounds is not vast.
1.3 Structures of Formazans
As previously stated formazans are usually characterized by their intense colours. They are commonly,
solids of a relatively low melting point when compared to the size of the molecules.
viii

Figure 6: 1,3,5-triphenyl formazan
Triphenyl formazans (illustrated in Figure 6) were often particularly soluble in acetone and chloroform.
However their solubility in water was negligible.
It was first thought by Pechmann and Runge that formazans were tautomeric, however their results
were inconclusive.3
In 1941, Hunter and Roberts had studied the proposed several pairs of formazans;
they had found that these pairs were identical. Therefore not existing as tautomeric formazans. Instead
what was suggested that the reason the formazans were seen as tautomeric was due to an internally
coordinated structure so that it forms resonance hybrids, where the ‘tautomeric’ imino- hydrogen forms
a chelated hydrogen-bridge structure (as shown in Figure 7).vi
Figure 7: Formazan mesomeric structures
ix

Formazans are good examples of organic and organometallic photochromism. With organometallic
being the more favoured and widely used, however the Hg2+
Dithizone complex ((1,5-diphenylthiocarbazone-N,S-)mercury(11)) is a well-used example of organic of
organic photo chromes.vii
Looking at the illustration shown in Figure 8:
Figure 8: Yellow solution of (1,5-diphenylthiocarbazone-N,S-)mercury(11) (left) and Blue Solution of ((1,5-
diphenylthiocarbazone-N,S-)mercury(11)) (right)7
These form again strongly coloured compounds, this reaction (Figure 8) is irradiation by visible light of
the solutions which then induces a reversible colour change from yellow to blue.viii
This is one of many
examples of how a slight change in the structure of a formazan can result in an intense change in
colour.
In another experiment conducted by Hausser in 1949, he showed a similar type of reaction of formazans
that changed from red to yellow following their exposure to visible light.ix
However a formazan molecule could be one of four possible structures (shown in Figure 9 below), this is
due to the geometrical isomerism about the two double bonds, C=N and N=N, the orientation about the
N=N bond is described using cis/trans and the orientation about the C=N bond is described using
syn/anti nomenclature.3
Figure 9: (from left to right) Cis-syn (chelates), Cis-anti, Trans-syn(chelates), Trans-anti
The cis-anti and trans-anti forms do not give a chelate structure because of the position of the N-H,
however the cis syn and trans-syn contain a chelate structure which involves hydrogen bonds, even
though the two form chelate structures the trans-syn is the most favoured, as the cis-syn has more steric
interfence.5
x

These formazans, which do not chelate (as shown above) are yellow, while molecules, which do chelate
and form the hydrogen bridge are red, this is due to the presence of two multiple bonds in the formazan
chain, the N5(denoted with an * in Figure 9) atom also has a lone pair electron that brings about the
formation of the overall conjugated system, see Figure 10 for illustration:5
Figure 10: Red formazan (left), chelate form Yellow formazan (right), open form
1.4 Tetrazolium salts
As previously discussed, Friese had synthesized the first formazan in 1875, while the first tetrazolium
compound was not synthesized until 1894 by von Pechmann and Runge. These tetrazolium salts contain
a ring consisting of one carbon atom and four nitrogen atoms; one of these nitrogen atoms will also be
quaternary (see Figures 11 and 12 for examples).
Figure 11: Tetrazolium ion derived from (2H) Tetrazole (left), (2H) Tetrazole (right)
These salts can also have resonance structures such as:
xi

Figure 12: Tetrazolium ions derived from (1H) tetrazole (left), (1H) Tetrazole
Most of the know tetrazolium salts as shown in (Figure 11) have been derived from (2H)tetrazole
(Figure 11). However the ions shown in Figure 12 are derived from (1H) tetrazole, while these are not
common they are theoretically possible.6
These tetrazoles have polynitrogen electron-rich planar structural features, due to these features;
tetrazole derivatives (tetrazolium ions) are very useful drugs, and explosives. Most notably they have
many applications in medicine, in this industry the functional group on the tetrazole was used as a
replacement for carboxylic acid, as the pKa was close and it also contained the same delocalized system
space requirements.
Due to the planar ring structure and the nitrogen-rich multi electron conjugated system, tetrazole
derivatives (tetrazolium ions) can have both acceptor and donor properties.x
Tetrazolium salts are usually colourless or faintly yellow, however when reduced they give deeply
coloured compounds (formazans). These salts are stable, crystalline and usually formed with weak
acids.5
A widely used tetrazolium reaction is the blue tetrazolium reaction for the analysis of
corticosteroids, blue tetrazolium (as shown in Figure 13 below) and triphenyltetrazolium (illustrated in
Figure 14) oxidize the alpha-keto moiety of the C17 side chain when in a strongly alkaline solution.
They are then reduced to highly coloured formazans.xi
Figure 13: Blue tetrazolium [3,3’-(3,3-dimethoxy -4,4-biphenylylene)bis(2,5-diphenyl-2H-tetrazolium chloride)]
xii

Figure 14: Triphenyltetrazolium (2,3,5-triphenyl-2H-tetrazolium chloride)
Another widely used tetrazolium salt is MTT (illustrated in Figure 15), which is used in measuring a
cells viability or proliferation rate in order to evaluate the cytotoxic activity of drugs. It has been used so
much due to the absorption maximum (λmax) of its formazan product (illustrated in Figure 15) is at a
longer wavelength (565nm) while it also is able to be reduced without an electron carrier.xii
Figure 15: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (left) and (E,Z)-5-(4,5-
dimethylthiazol-2-yl)-1,3-diphenylformazan (Formazan) (right)
1.5 Synthesis of Tetrazolium salts and Formazans
There any many methods proposed for the synthesis of formazans, the first being the condensation of
aromatic and aliphatic aldehydes with phenylhydrazine, (reaction illustrated below in Figure 16) and
the coupling reaction of the subsequent hydrazones with diazonium salts. These were made to try and
synthesize a less toxic formazan to be used in the medical industry as the previous formazans know
where toxic in nature.xiii
xiii

Figure 16: The synthesis of Formazans by reaction of diazonium salts on arylhydrazones
Most of the formazans know today are synthesized by this method.
The second synthesis is that of modifications of substituents, which are present in formazans. In this
method formazans may be converted to other formazans by inducing changes in the functional groups,
which are substituted in the molecule.
This is achieved by the hydrolysis of ester and nitrile substituents to the carboxylic acids and of N-acyl
groups to free amino groups (these have been reported in particular cases). The carboxylic acids have
been esterified through silver salts.5,6
The third synthesis is by the reaction of diazonium salts, in which the salts are coupled to many
different compounds containing active methylene groups, such as aldehydes, ketones, nitroalkanes,
malonic esters etc.6
When the substitution of one diazonium ion in the methylene group is achieved, the
result is an azo compound, this then rearranges to a phenylhydrazone. When two molecules of a
diazonium salt are used, the second diazonium ion couples to the phenylhydrazone molecule, which was
formed beforehand.5
An example of this method were the preparation of macrocyclic crown-formazans,
these were prepared with pyruvic and arlypruvic acids (illustrated in Figure 17).xiv
The final proposed synthesis of formazans was by the reduction of tetrazolium salts, however
tetrazolium salts are prepared by the oxidation of formazan compounds, therefore this method is
negligible.
xiv

Figure 17: Synthesis of Macrocylic crown-fromazan5
Tetrazolium salts are prepared by the oxidation of formazans, in previous years various oxidants have
been used such as mercuric oxide, nitric acid, isoamylnitrile, halogenamides etc.6
Different oxidants
have been used under various conditions and therefore it is difficult to know which is the best reagent
to use in each given case. However, all mechanisms occur in a similar way and at the same time by
dehydrogenation and cyclization. In an experimental environment the easiest way to determine whether
the oxidation reaction is complete is the disappearance of the intense formazan colour. However in this
project we will be looking for the intense colour returning to the solution therefore showing the
chemical warfare agent.
xv

2. Experimental
2.1 Experiment 1:
This experiment that was conducted by forst building setting up a spectrometer, as shown in Figure 18:
Figure 18: Diagram of built spectrometer
In this experiment a spectrometer was set up in the form above, a light source (has to have a constant
light intensity throughout), held in place by clamp stands. Above the light source a filter (which can be
moved back and forth into place in order to determine the absorbance) was fixed, and a pinhole, which
is attached to a micrometer, the position of this pin hole, the visible spectrum itself is produced by the
grating and the micrometer adjusted in order to select certain regions of wavelengths (these have
bandwidth).
Two photographs of the spectrum were taken at each micrometer position, one without the filter, the
second with. The micrometer positions taken were as follows:
0m-6
, 0.2m-6
, 0.4m-6
, 0.6m-6
, 0.8m-6
, 1m-6
, 1.2m-6
, 1.4m-6
, 1.6m-6
, 1.8m-6
Using this method, however with different filters of a Red 610nm long pass, an Orange 550nm long
pass, a green 495nm long pass and a blue filter were used. The exposure of light on the detector was
lowered in order to enable easier detection.
xvi

The photos were then analyzed using Matlab, as the RGB values alone are not a sufficient test of
wavelength, thus the images are converted to greyscale in order to obtain light intensities and
absorption.
2.2 Experiment 2:
Using the same method as the previous experiments, the same filters were used, however using data
obtained from previous experiments; photographs were taken every 4m-6
about the absorbance point.
For each filter the micrometer positions used were:
Red 610nm: 0.2m-6
, 0.24m-6
, 0.28m-6
, 0.32m-6
, 0.36m-6
, 0.4m-6
, 0.44m-6
, 0.48m-6
, 0.52m-6
,
0.56m-6
, 0.6m-6
Orange 550nm: 0.6m-6
, 0.64m-6
, 0.68m-6
, 0.72m-6
, 0.76m-6
, 0.8m-6
, 0.84m-6
, 0.88m-6
, 0.92m-6
,
0.96m-6
, 1m-6
Green 495nm: 1.2m-6
, 1.24m-6
, 1.28m-6
, 1.32m-6
, 1.36m-6
, 1.4m-6
Blue 350-500nm: 1.2m-6
, 1.24m-6
, 1.28m-6
, 1.32m-6
, 1.36m-6
, 1.4m-6
2.3 Experiment 3:
Modifying the spectrometer in order to house the cuvette and using the same method as previous
experiments, four separate cuvettes were analyzed. Each cuvette was filled with 1 cm of solution, apart
the first being a blank cuvette, the second was of full concentration of formazan, the third at half
dilution and the fourth at three quarters dilution. The micrometer positions used were:
0m-6
, 0.2m-6
, 0.4m-6
, 0.6m-6
, 0.8m-6
, 1m-6
, 1.2m-6
, 1.4m-6
, 1.6m-6
, 1.8m-6
3. Results and Discussion
The photos of the grating taken from the detector, which compared light intensity without a filter and
with a filter, were put into the MatLab program and analyzed. The images were converted from RGB to
Greyscale in order to determine the light intensity of the image through the grating.
xvii

The spectrometer built to carry out these experiments was constructed to be as compact as possible with
the equipment available, while also being capable of wavelength differentiation (which it is). However
before analyzing the actual formazan complexes the micrometer positions had to be assigned to
wavelengths. This was achieved by standardizing the spectrometer with already calibrated filters. The
results are given below.
The first filter analyzed was the Red 610nm long pass; the results of this are shown in table 1. This
filter would only allow light with a high wavelength, above 610nm, to pass through. To determine the
absorbance of the filter (in later experiments this would represent the warfare agent solution) the
intensity without a filter was taken also. Using these intensities the absorbance was able to be calculated
using the equation:
The graphical representation is shown in Figure 19.
The absorbance is represented in arbitrary units in order to show the ratio of intensities.
3.1 Red 610nm long pass filter
Table 1
Micrometer
Position
Intensity of light W/O
filter
Intensity of light
W/filter
Absorbance from
filter
0 95.1744 87.8974 0.034544119
0.2 115.4408 86.9342 0.123168666
0.4 142.2336 55.2434 0.410721803
0.6 189.2781 44.9749 0.624130161
0.8 151.9173 11.6391 1.115687834
1 155.2679 0.4524 2.535559083
1.2 126.977 0.2632 2.683439177
1.4 59.6728 0.2941 2.307281392
1.6 58.2708 0.0625 2.969570963
1.8 23.0181 0.2846 1.907834577
xviii

Figure 19: Absorption of 610nm filter
Figure 19 shows that the red filter absorbs light strongly at 0.6m-6
. This was as expected as the Red
filter is at 610nm so therefore will absorb a considerable amount of light. This filter did not absorb as
strongly as expected as it has more gradual increase in absorption, this would be due to other
wavelengths contaminating the image and therefore affecting analysis.
3.2 Orange 550nm long pass filter
The second filter used was an orange 550nm long pass filter; the results of this experiment are shown in
table 2, while the graphical representations are shown in Figure 20.
Table 2
Micrometer
Position
Intensity of light
without filter
Intensity of light with
Orange filter
Absorbance from filter
0 79.2684 96.1801 -0.083985132
0.2 173.3711 154.5576 0.049886339
0.4 203.1176 207.4372 -0.009139085
0.6 182.2536 205.1244 -0.051341208
0.8 133.7148 188.6367 -0.149446711
1 161.0703 118.8367 0.132064884
1.2 163.1837 55.8209 0.465879942
1.4 122.2599 1.0526 2.065020671
1.6 59.1172 0.6648 1.949022845
1.8 56.2912 0.7582 1.870656727
xix

When looking at Figure 20, the observation is that the filter begins to absorb strongly at the 1m -6
. The
absorption of this filter is very strong which shows that the spectrometer is providing results, which
were expected.
This absorption from the orange 550nm long pass filter was predicted as it would let larger wavelengths
pass through, (absorbing less than the 610nm filter) which would correspond close to the orange to red
end of the spectrum (590nm-780nm), while absorbing light, which is closer to the blue end of spectrum
due to the shorter wavelengths (390nm-500nm).
From these experiments it can be determined that the wavelength corresponds to micrometer positions:
610nm = between 0.2 and 0.6 micrometer position
550nm = between 0.6 and 1 micrometer position
In order to determine the wavelengths in accordance to all of the micrometer positions, additional filters
were used with lower wavelengths than the previous.
3.3 Green 495nm long pass filter
Firstly a Green 495nm Long Pass filter was fitted and the experiment repeated as before. The results are
shown in table 3, while the graphical representation of the absorbance is shown in Figure 21.
Table 3
xx

Looking at Figure 21 it is observed that the strong absorbance of 495nm is between the micrometer
positions 1.2m-6
and 1.6m-6
. Again the absorption of this filter is very definitive and strong, showing
that the spectrometer is working accurately.
3.4 Blue Filter
To account for the final blue region of the spectra (which is observed from 400nm – 500nm) a blue
filter was used. Unlike the previous filters this had not been calibrated to an exact absorbance, however
by looking at a UV spectrum of the sample it was deduced that it had an absorbance between the
regions 350nm-450nm. The results are shown in table 4 while the graphical representation is shown in
Figure 22.
Micrometer
Position
Intensity of light
W/O filter
Intensity of light
W/filter
Absorbance
from filter
0 76.4342 72.8627 0.020782464
0.2 116.2653 110.1814 0.02334183
0.4 122.712 116.8322 0.02132448
0.6 166.8866 162.3288 0.012025889
0.8 136.0317 126.0068 0.033246143
1 155.8685 150.0998 0.016378242
1.2 138.7128 136.7687 0.006129819
1.4 89.9569 65.9569 0.134774245
1.6 61.746 1.7959 1.536326678
1.8 49.1859 0.8753 1.749683694
xxi

Table 4
Figure 22: Absorption of blue filter
Looking at Figure 22 it demonstrates a large absorbance over the spectrum, firstly absorbing notably
between 1.2m-6
and 1.4m-6
. This region was previously discussed when using the green filter, yet the
results were expected as the green filter begins absorbing at 495nm, while the filter used in this
experiment seemed to absorb most of the light which was in the blue region (500nm) thus there would
be a crossover and absorbance’s would be of a similar region. There is also another peak in absorbance
between 0.2m-6
and 0.4m-6
. This was observed, as the filter used is a short pass filter meaning that it
Micrometer Position
Intensity of light W/O
filter
Intensity of light
W/filter
Absorbance from
filter
0 94.7551 1.7868 1.72452665
0.2 133.5601 3.1655 1.625234418
0.4 170.0136 13.7664 1.091663279
0.6 146 15.4308 0.975964413
0.8 157.3515 22.3197 0.848182534
1 156.5238 32.7642 0.679180815
1.2 114.9365 59.3129 0.287308809
1.4 61.0794 58.5374 0.018461333
1.6 64.6555 60.2812 0.030423585
1.8 55.2132 53.3651 0.014785591
xxii

light with a shorter wavelength will be detected on the CMOS/CDD camera, and between these two
points the presence of light with wavelength which corresponds to the short pass filter was seen,
meaning that a small amount of intensity was identified on the camera.
In order to determine the wavelengths more accurately the same filters were used and the experiment
repeated however with smaller micrometer alterations in the ranges of strong absorption stated above
for each filter.
3.5 Red 610nm long pass filter (precise measurements)
Following this the Red 610nm long pass filter was analyzed using smaller micrometer alterations of
0.04m-6
. The recordings started at 0.2m-6
as this was when the filter had began absorbing strongly. The
results are shown in table 5 and the graphical representation in Figure 23.
Table 5
Micrometer Position
Intensity of light
W/O filter
Intensity of light
W/filter
Absorbance from
filter / AU
0.2 133.1791 76.2889 0.241974723
0.24 132.2215 81.7395 0.208870103
0.28 149.4172 78.8005 0.277871621
0.32 157.2971 79.4943 0.296384726
0.36 164.2585 78.4218 0.321091046
0.4 158.3628 76.5805 0.315534974
0.44 169.6576 69.6576 0.386604812
0.48 180.0272 69.2336 0.415021212
0.52 165.5964 62.0544 0.42627831
0.56 155.2494 39.4467 0.595019253
0.6 164.7596 36.7438 0.65166666
xxiii

Figure 23 shows the absorbance of the Red 610nm filter (by box analysis), (strong absorbance at 0.4m-6
)
however there is not a sharp absorbance as the Orange 550nm filter had produced and what was
expected. This may have been due to when taking the intensities the micrometer movements were so
small that a red wavelength was still present (a reduced intensity each time), thus providing difficulty
when analyzing in Matlab. Also the images would have been contaminated with other wavelengths
through the pinhole.
3.6 Orange 550nm long pass filter (precise measurements)
The Orange 550nm long pass filter’s results are shown in table 6, while the graphical representation
Figure 25.
Table 6
Micrometer
Position
Intensity of light
W/O filter Intensity of light W/filter
Absorbance from
filter / AU
0.6 186.7175 181.2899 0.012811414
0.64 194.3469 182.254 1.066351904
0.68 211.7914 206.6485 1.024887188
0.72 216.125 202.2344 1.068685644
0.76 205.7392 200.7392 1.02490794
0.8 175.263 174.0998 1.006681225
0.84 166.2494 163.2472 1.018390515
0.88 163.7147 157.0194 1.042639954
0.92 152.0923 122.932 1.237206748
0.96 221.6667 147.9388 1.498367568
1 224.8032 155.5778 1.4449568
xxiv

By observing Figure 25 it is seen that the filter begins strongly absorbing between 0.88m-6
and 0.92m-6
,
so therefore the wavelength between these solutions is 550nm.
Again this absorption of this filter is very intense presenting the accuracy of the spectrometer.
In order to gain full wavelengths aligned to the micrometer, further analysis was needed in order to
gain a rough wavelength for the full visible light spectrum. This was desirable before analysis of actual
solutions of the pseudo chemical warfare formazan.
3.7 Green 495nm long pass filter (precise measurements)
Following on a Green 495nm Long Pass filter was fitted and the precise experiment repeated as before.
However beginning at 1.2m-6
. The results are shown in Figure 26.
From Figure 27 it can be examined that the absorbance of 495nm is between 1.32m-6
and 1.36m-6
. Much
like the orange 550nm long pass before this again shows a very strong absorbance.
3.8 Blue filter (precise measurements)
The results for the blue filter are shown in Figure 27.
xxv

Analyzing Figure 27 it can be seen that there is a strong absorbance from 1.36m-6
to 1.4m-6
, therefore it
is safe to assume that the micrometer positions are between 350nm and 500nm. Due to this filter not
being calibrated, in order to improve the accuracy of the wavelengths at the micrometer positions these
results were compared to that of the green filter (when the micrometer positions were more precise).
3.9 Assignment of wavelengths to micrometer positions
When comparing this to Figure 26 and the accurate measurements calculated previously, a final precise
spectrum over the micrometer positions was projected:
Blue wavelength region 350nm – 495nm = 0m-6
– 1.4m-6
495nm = 1.32m-6
– 1.36m-6
550nm = 0.84m-6
– 0.92m-6
Red wavelength region 610nm = 0.4m-6
– 0.44m-6
From these results it can be determined that the compact built spectrometer appears to be able to
differentiate between wavelengths reasonably reliably.
Proceeding from these calculations, solutions of the formazan compound, which was formed by reaction
with the pseudo warfare agent, were ready to be analyzed. Thus the original spectrometer that was built
needed to be modified in order to house the cuvettes, which contained the formazan. Subsequently when
this was accomplished, the results of an empty cuvette were taken to account for the absorbance of the
cuvette; these results were taken into account when analyzing the solutions.
xxvi

3.91
Analysis of Formazan complexes
From the results of the Red 610nm and Orange 550nm long pass filter it is expected that the non-
diluted formazan complex to be analyzed would absorb near the 550nm – 610nm region, due to the
similar colour the complex.
The formazan complexes do not have concentration measurements as, since they were formed from the
reduction of the tetrazolium solution (1mg per ml), it cannot be known the exact concentration.
Therefore they have been labeled as:
Non diluted, dilution of factor 2 and dilution of factor 4
The purpose of this experiment was a sensitivity test of the spectrometer with respect to the formazan
complex. This meant identifying at what dilution the spectrometer would cease being accurate and
therefore not usable.
The first formazan compound analyzed was a non-diluted solution of full concentration; 1cm of solution
was added to the cuvette. The second was the original formazan complex with a dilution factor of 2,
while the third was the original formazan complex with a dilution factor of 4. The results of this are
shown in Figure 28, 29 and 30 respectively.
xxvii

Figure 28: Absorption of Non-diluted formazan complex
Figure 29: Absorption of Formazan complex at dilution factor 2
Figure 30: Absorption of formazan complex at dilution factor 4
Firstly looking at Figure 28, it can be seen that there is a strong absorbance 1m-6
to 1.2m-6
. This
absorption is very intense, however it absorbs at a slightly lower wavelength (micrometer position 1-
1.2m-6
, expected = below 1). However the results are still promising, for this solution.
xxviii

However due to the fluctuating patterns of the graphs of the analyzed solutions, a new method to
improve the results was approached.
3.92 Boxed Analysis of Formazan complexes
In the technique used previously the filters and solutions had been examined by choosing the area of the
image, which contained the largest light intensity. However the new method approached was to use a
35x30 pixel ‘box’. In each image the ‘box’ was moved across the spectrum by 30 pixels a time and an
average intensity was taken. This would increase the accuracy of the spectrometer as it takes into
account the regions of light, while accounting for the contaminating wavelengths, which are seen
through the pinhole at certain micrometer positions. Therefore improving wavelength resolution. This
is illustrated in the diagram in Figure 31:
Figure 31 shows the boxed analysis
method, by looking at the image it is
observed that a number of wavelengths are contaminating the spectrum, therefore by moving said box
along the spectrum (30 pixels a time), the intensities measured are improved
The results compared to their analogous in the previous method are shown below in figure 32.
xxix
Figure 31: Box analysis of standard spectrum, the arrow
represents a movement of 30 pixels
Figure 32: representing absorbance of solutions

By looking at Figure 32, it is easy to see a large spike in the absorbance at 1m-6
for the boxed analysis
and 1.2m-6
for the original analysis. Therefore it is safe to assume that the boxed analysis is the most
accurate absorbance due to the averaging of the intensities.
Figure 33 shows the comparison of the boxed analysis with the original spectroscopic method, the
results of both shows the solution begins to start absorbing strongest at 1.2m-6
. However before the box
analysis there were great fluctuations of absorbance between 0.4m-6
and 1m-6
as it drops from 0.234 to
0.005, while it would be expected that this region between these values would be relatively similar up
xxx
Figure 33: representing absorbances of solutions

until the solution would start absorbing greatly at a 1.2m-6
. However in the boxed analysis graph there is
a comparable pattern of a dip in absorbance however it is only from 0.22 to 0.12.
Figure 34 shows the comparison of the formazan complex at dilution factor 4. While the original
analysis shows major fluctuations of absorbance, it would have to be assumed that its major absorbance
beings at 1.6m-6
. However if this is compared to the box analysis, which has been determined to be more
reliable, it can be said that the actual absorbance begins at 1.2m-6
, as this is the largest spike in
absorbance. This too would mirror the half diluted formazan complex, as this begins absorbing at 1.2m-
6
. The absorbances at this point are also similar in terms of their absorbance intensity:
Dilution factor 2 (F1): 1.2m-6
= 0.13
Dilution factor 4 (F2): 1.2m-6
= 0.12
However F1 complexes absorb maximum is a considerable amount greater than F2’s, which is to be
expected.
Thus referring back to the previous micrometer calculations, it is observed that the absorbance is at
wavelength:
Non-diluted solution = 495nm – 550nm
Secondly observing Figure 33, it can be seen that this solution begins absorbing at 1.2m-6
to 1.4m-6
,
slightly further along than the non-diluted solution, it also absorbs at a less intensity than the non-
diluted solution (which is expected). When comparing these results to the micrometer positions
calculated, the absorbance is at wavelength:
Dilution Factor 2 solution = 400nm – 495nm
Lastly in Figure 34 the Quarter-diluted solution’s absorbance is not as clearly defined as the two
solutions previously, even with the box analysis. Therefore it can be concluded that the sensitivity of the
compact spectrometer would terminate at a dilution factor of 4.
xxxi
Figure 34: Representing absorbance of the solutions

These results were as expected due to the absorbance becoming less intense as the dilution factor
increased as well as the absorbance of the wavelengths becoming shorter as the complexes were diluted.
3.93 Analysis of colour change using coded image analysis
In addition to the spectroscopy, the second approach to creating a detector was a coded program, in
order to distinguish the slightest colour change and inform the user if an antidote was needed. This self-
program is at present unfinished, however this program was able to load a given image of a solution
such as Figure 35 below:
Figure 35: Example of image used in program
xxxii

The program would then convert the image from various ranges such as .pdf, .jpeg, grayscale etc. to a
RGB image. The image would then be split into three separate bands of red, green and blue and the
separate images displayed (Figure 36).
Figure 36: (from left to right) Red band image, Green band image and Blue band image
The following part of the program puts each of these split bands into different arrays in order for their
histograms to be calculated. The program would then show the histogram chosen; in this case the red
histogram was displayed due to the colour of the solution (shown in Figure 37):
Figure 37: Histogram showing red pixel values over the image
Once this part of the program had run, the second part of the program needed to be defined. This part of
the program analyzed the colours in the image itself through clustering. The clustering was achieved by
finding parts of the image, where there were a large cluster of pixels with similar or the same RGB
xxxiii

values. This would improve the detection, as the program would be able to detect the smallest amount of
clusters. The number of clusters could be decided by the user.
The following part of the code was proceeded using ‘kmeans’. This would search entire image of the
users choice and group each individual pixel by its RGB value, these RGB components would be
evaluated and the most dominant colours found (the number of clusters is chosen by the user). With
these dominant colours a threshold was written into the program, which would be sensitive to the
relative amount of red compared to the green and blue values of the dominant colours.
After the program had produced an image it would then ask the user:
Please input the factor you would like to use (how many times larger must
R be than GB)
For example if the user expected a minuet amount of the warfare agent, he/she would input the
sensitivity at a low value e.g. 2, however if searching in general, the user would use a larger sensitivity
e.g. 7.
xxxiv

4. Conclusion
In conclusion the aims of this project have been met as a sensor to enable the detection of a formazan
complex. This has been achieved in two forms, the first being by calibrating different micrometer
positions, which correspond to different wavelengths. Then by monitoring the absorption of
wavelengths, in relation to the micrometer, of different dilutions of a pseudo warfare treated formazan
complex.
The second being a coded program that analyzes clusters of RGB values of a given image, this would be
beneficial, as it would enable detection of an extremely faint colour change. Due to the difficulty of
detecting small colour changes over time with eye detection alone. This program would be very
advantageous. Also as seen in the results section above, as the dilution of the complexes increase it can
become very difficult to distinguish at what wavelength the complex is absorbing at. Using said
program instead to detect a slight colour change may give a better result at certain dilutions.
When building the spectrometer in this project, the simpler, smaller and ease of use the better, however
this did have drawbacks, as selectivity was not accounted for even though the spectra have clear
absorbance’s, the absorptions of the solutions are not as accurate and intense as what was expected.
Moving forward with this project I would attempt to make the spectrometer used smaller, this could be
achieved in various ways.
The first being replacing the micrometer adjuster, which was a bulky component, with a ‘SQUIGGLE
motor’ this is only 1.8x1.8x6mm. This would greatly decrease the size, while also increasing the
accuracy, as it would be programmed to move the micrometer by exact units with no human error. An
LED light could also be used instead of a larger light, again decreasing the size. However would have to
obtain an LED that had a high light intensity. Also instead of using cuvettes to house the solutions
during analysis, which are tall when compared to the size of the spectrometer and the amount of
solution, glass petri dishes could be developed, which would also improve ease of use as they could be
slid over the LED light relatively easily.
I would also recommend to finishing the original coding program. This would entail writing code,
which after the clusters had been obtained and shown, could show the center of each cluster on the
image (centroiding). Once this had been completed, the next step would be to find the LD50 of each
warfare agent and to treat these to the swab test or the tetrazolium in order to induce the formazan
complex. These solutions would be analyzed using the built spectrometer in order to determine the
specific wavelength they would absorb at. While also taking photos of the solutions themselves and
running them through the coded program.
Moreover in the future I would consider using both these methods in conjunction with one another to
ensure the most accurate of results. This was needed in some cases when using the spectrometer, as
clustering (such as the code used in the program) could have been used when moving the micrometer
small distances. For example when the full concentration of formazan were being analyzed, when it had
began to absorb some of the colours shown when the solution was not present, the lowered intensity was
xxxv

taken into account due to the larger colours not being absorbed giving a larger intensity. To combat
this, the program coded could be used to identify the certain clusters of light; the light intensities of the
different clusters could then be calculated and averaged, in order to give a more accurate absorption of
the solution. This problem was partly solved when using the ‘boxed analysis’ method, however this
proved time consuming as a user. Therefore writing a piece of code into the program, which could
provide this analysis for the user, would be definitive in the success of the final product.
Finally these two methods work differently as the spectrometer relies on the grating to produce the
wavelength, while the program relies on detection of certain RGB values of pixels. In the future these
two could be used together in order to finish developing the sensitive, cheap and portable sensor.
5. References
xxxvi

i
F. Kanal, D. Schleier and P.Nuernberger, ChemPubSoc Europe, 15, 3143-3146
ii
I. Fallis, Highly Specific Colormetric Chemistries for Chemical Warfare Agents, 2015
iii
Yusra H Al-Araji, Jawad K Shneiene* and Ahmed A Ahmed, International Journal of Research In
Pharmacy and Chemistry, 2015, 2231-2781, 41.
iv
Altam FP, Histochem. Cytochem. 1976, 9 (3): 1-56.
v
H. Senoz, Hacettepe J. Biol & Chem. 2012, 40(3), 293-301
vi
L. Hunter and C. Roberts, Associating Effect of the Hydrogen Atom. Part IX. The N-H-N bond.
Virtual tautomerism of the Formazyl compounds. 1941, J. Chem. Soc, 820-823
vii
G. R. Burns and C. W. Cunningham, Photochromic Formazans: Raman Spectra, X-Ray Crystal
Structures, and 13
C Magnetic Resonance Spectra of the Orange and Red Isomers of 3-Ethyl-1,5-
diphenylformazan, 1988, J. Chem. Soc, 1275-1280
viii
N. L. Cromhout and A. T. Hutton, Applied Organometallic Chemistry, 2000, 14, 66-74
ix
L. Hausser, D. Jerchel and R. Kuhn. Red-Yellow Rearangement of Formazans by Light. R Chem
Ber, 1949, 82, 515.
x
C. Wei, M. Bian and G. Gong, Molecules, 2015, 20, 5528-5553
xi
R. E. Graham, E. R. Biehl and C. T. Kenner, Journal of Pharmaceutical Sciences, 1977, 66, 965-
970.
xii
M. Ishyama, M. Shiga, K. Sasamoto, M. Mizoguchi and P. He, Chem. Pharm, Bull., 1993, 41,
1118-1122.
xiii
H. Tezcan, E. Uzluk, Science direct: Dyes and Pigments, 2007, 75, 633-640
xiv
Y. A. Ibrahim, A. H. M. Elwahy and A. Abbas, Tetrahedron, 39, 11489-11498

Dissertation

More Related Content

What's hot (6)

Viewers also liked (20)

Similar to Dissertation (20)

Dissertation