Method development for quality control of the primary explosive, Potassium 4,6-Dinitrobenzofuroxan (KDNBF)

Method development for quality control of the primary explosive,

Potassium 4,6-Dinitrobenzofuroxan (KDNBF)

Edvin Elmroth Master thesis 30 hp (30 credits) 2020-05-19 Supervisors: Sjöberg, Viktor; Persson, Fredrik; Karami, Pishtiwan Examiner: Karlsson, Stefan

Abstract

“Green” explosives are an important sub-family of explosives due to the banning of explosives based on heavy metals, such as lead azide and lead styphnate, according to the REACH list. A substitute to lead azide is potassium

4,6-dinitrobenzofuroxane (KDNBF) which is an alternative “green” explosive. Hence there is an upcoming need for analytical and quality control protocols for KDNBF. In this report is HPLC-UV/VIS, GC-FID and potentiometric titration evaluated for their suitability as methods for testing of KDNBF. The results implies that KDNBF can be analyzed by GC-FID, when dissolved in DMSO and caffeine is used as internal standard. To analyze KDNBF by HPLC-UV/VIS was isocratic mode used together with a porous graphitic carbon (PGC) (Hypercarb ®) column. However, degradation of the analyte was severe and quantitative results were not

obtained. Potentiometric titration indicated that KDNBF equilibrate with four hydrogen ions during titration, instead of the one to one ratio with potassium and hydrogen ion, as in previously proposed theory. Despite this and with careful titration is potentiometric titration and GC-FID analysis combined suitable as a quality control protocol for KDNBF.

Keywords

Table of Contents

Abstract ... 2

Keywords ... 2

Introduction ... 5

Deflagration and detonation ... 6

Deflagration ... 6 Detonation... 6 Explosive properties ... 7 “Green” explosives ... 8 REACH ... 8 Toxicity ... 8 Lead ... 8 Silver ... 8 Potassium 4,6-Dinitrobenzofuroxan ... 9

Instrumentation (settings and performance) ... 9

Gas Chromatograph (GC) ... 9

Liquid chromatograph (LC) ... 9

Titration ... 9

Aim ... 11

Materials and Methods ... 11

Materials ... 11 Chemicals ... 11 Labware ... 11 Methods ... 12 Titration ... 12 Metal analysis ... 12 GC-FID ... 12 HPLC-UV/Vis ... 12 Instrument ... 13 Protection ... 13 Scale ... 13 pH – Meter ... 13 AAS ... 13 GC-FID ... 13 HPLC-UV/Vis ... 14

Results and discussion ... 16

Observations ... 16

Titration... 16

Metal analysis ... 18

GC-FID ... 19

Conclusions ... 22

Acknowledgements ... 22

References ... 23

Introduction

A current topic within many businesses in Europe is to supply the market with environmental friendly products, according to environmental laws and

agreements within the European Union (Europaparlamentets och rådets

förordning (EG) nr 1907/2006, 2006; European Parlament and the council of the european union, 2013) Products that are environmentally friendly are hence becoming more and more important because of stricter regulations and laws (Europaparlamentets och rådets förordning (EG) nr 1907/2006, 2006). Within the field of explosives, there are possibilities to make the explosives more

environmental friendly. This can primarily be done by substituting the traditional heavy metal-based explosives, such as lead styphnate, lead azide, silver azide, mercury fulminate, cadmium azide, cobalt nitrotetrazole (Matyáš & Pachman, 2013) etc. with “green” explosives that contain less environmentally unfriendly elements (Tchounwou, Yedjou, Patlolla, & Sutton, 2012; Ilyushin & Shugalei, 2019).

Explosives

The history of explosives starts in China approximately one thousand years ago with the invention of primitive predecessors to black powder. The first explosives were composed of products containing coal, sulphur and potassium nitrate (Sejlitz, 2014). Almost a millennium later came the next major breakthrough with the invention of nitroglycerin and the ignition cap, patented by Alfred Nobel in 1864-66 (Larsson, 2010). Modern characterization of explosives has since then been divided into four major groups, with different characteristics and purposes: propellants, primary explosives, secondary explosives and pyrotechnical

mixtures as described below (Sejlitz, 2014).

Propellants

The primary function of propellants is to create pressure that sends a projectile in a predicted direction. Propellants can be single based, double based or multi based depending on the number of different types of explosives that the

propellant is composed of (Sejlitz, 2014). Single based propellants are based on nitroglycerin and nitrocellulose, in reality two explosives, but is of tradition still considered as single based. Double based and multi based propellants consist of two or more components to enhance different properties of the propellant such as pressure increase against burn time etc. (Sejlitz, 2014).

Primary Explosives

Primary explosives are used to transform e.g. a mechanic, thermic or electric input to another compartment via combustion, hence is the major function for this group of explosives to be sensitive for the mention inputs (Matyáš & Pachman, 2013). There are different compositions of primary explosives such as;

fulminates, azides, organic peroxides, tetrazoles as well as salts with metallic cations and organic anions, such as benzofuroxanes or polynitropheoles (Matyáš & Pachman, 2013).

Secondary Explosives

The purpose of secondary explosives is to detonate once they reach a designated target. This is done via ignition by a primary explosive that has a deflagration to detonation transition (DDT) (Sejlitz, 2014). A DDT is necessary to ignite the secondary explosive (Mcfee, 2010), an example is dynamite burning openly or undergoing detonation by ignition from an ignition cap (Norberg, 2020). Famous secondary explosives include dynamite, nitrocellulose, nitroglycerin or trotyl (TNT) (Sejlitz, 2014). Secondary explosives are usually rich in -NO2,

-ONO2, and -NHNO2 groups, that form different nitrous-oxygen gasses when

detonated, but like all organic matter once combusted also water and different carbous-oxygene gasses (Norberg, 2020).

Pyrotechnical Compositions

Pyrotechnical compositions have a wide variety of different usages, such as delay effects, smoke formation, generation of highly intense lights etc.

Pyrotechnical compositions are also crucial elements in many other explosives, due to their ability to provide explosives with highly specific properties (Sejlitz, 2014; Norberg, 2020). Pyrotechnical compositions are usually powders where the different components are mixed rigorously through a sieve, and are mixed rigorous. One example is the photo flash where fine grained magnesium powder is mixed with potassium nitrate and zirconium stearate to create a bright white flash of light when ignited (Bennett, 1939; Norberg, 2020)

Deflagration and detonation

There are two important terms regarding the decomposition of explosives, deflagration and detonation. They can appear to be the same, in regard of exploding, but the main difference is how the surroundings are affected.

Deflagration

Deflagration is defined as a sub-sonic combustion process. If carried out in a container it will not be damaged, e.g. in a crucible will only the lid shoot off, once the explosive is ignited (Sejlitz, 2014). An example of such a product and

process is a cartridge for small fire arms, where black powder ignites inside the cartridge and generates enough gas for the bullet to be pushed out, without damaging the cartridge and weapon.

Detonation

The detonation process is defined by its very rapid combustion, with speeds in the range of 1-10 km s-1. In this application, the combustion is ignited via the

shockwave that travels through the material (Agrawal, 2011). In contrast to deflagration will a substance that detonates destroy its container, or is supposed to destroy it, instead of just shooting off the lid, (Sejlitz, 2014). Furthermore, it should be noted that certain energetic materials can undergo both deflagration and detonation depending on the conditions e.g. propellants can detonate under confinement (2.10.1 Properties and Behavior of Explosives, 2020; Norberg, 2020).

Explosive properties

Lead is used in explosives mostly in its cationic form. When producing lead based explosives is the predominant starting compound lead nitrate (Pb(NO3)2)

(Sejlitz, 2014). One of the more well-known lead based primary explosives is lead azide (Pb(N3)2) (LA) (Sejlitz, 2014). LA is relatively sensitive to friction and

electrostatic discharges (ESD), and can under normal conditions only detonate (2670 - 5500 m s-1) which is a desired property of a primary explosive (Matyáš &

Pachman, 2013). During the detonation nitrogen gas (N2(g)) and elemental lead

will form. Being a detonation process the products are scattered around on the detonation site (see Equation 1) (Matyáš & Pachman, 2013; Sejlitz, 2014).

𝑃𝑏(𝑁₃)₂→ 𝑃𝑏 + 3𝑁𝐸 ₂+ 443𝑘𝐽 (Eq. 1)

As a primary explosive, silver azide (AgN3) (SA) is commonly used (Matyáš &

Pachman, 2013). SA has a slightly slower detonation velocity than LA and it ranges between 1000 and 5000 m s-1. Compared to LA, SA is less sensitive to

impact test and friction test, see Table 1. However, SA it is more sensitive to ESD as seen in Table 1 (Matyáš & Pachman, 2013). Similar to LA, SA forms elemental silver (Ag) and nitrogen gas (N2(g)) during the detonation (see

Equation 2) (Matyáš & Pachman, 2013; Sejlitz, 2014).

An impact test is conducted by having a weight (fall hammer) dropped from different heights on a fixed amount of sample. From this, the h50 cm is

determined, meaning the height where 50 % of the test material detonates. A friction test is conducted by putting the sample between emery papers. The speed of dragging one of the emery papers is varied, and from this the amount of friction, in m s-1, that is needed to ignite 50 % is determined (Matyáš & Pachman,

2013; Norberg, 2020).

2𝐴𝑔𝑁₃→ 2𝐴𝑔 + 3𝑁𝐸 ₂+ 621𝑘𝐽 (Eq. 2)

Potassium 4,6-Dinitrobenzofuroxan (KDNBF) has similar properties as the previous mentioned azides except for the ESD test where it is approximately 103

times less sensitive for ESD, than LA and SA, see Table 1 (Mehilal, Suman, & Nirmala, 2002). The sensitivity and friction sensitivity for KDNBF, result in a limit of 20 mg to be handled safely.

Table 1. Data of explosive properties for lead azide, silver azide and KDNBF.

Lead azide

Silver azide

KDNBF

h50 cm (height where 50% detonates) 19.0 47.4 35.0

friction test (50% probability to ignite with emery paper m s-1)

1.2 2.6 3.8

“Green” explosives

REACH

The European Union has established the “Registration, Authorization and

Restriction of Chemicals” (REACH) with the purpose of addressing the chemical

usage, and its effect on the environment and humans in the European Union (Europaparlamentets och rådets förordning (EG) nr 1907/2006, 2006). As a consequence of REACH there is a constant and necessary strive towards minimizing the amount of heavy metals reaching the environment from anthropogenic sources. Based on the toxic effects of both lead and silver, on humans as well as on the environment, there is consequently a need for environmentally-friendly substitutes in the field of explosives (Tchounwou, Yedjou, Patlolla, & Sutton, 2012). Within REACH is LA listed under annex XVII as a chemical causing e.g. reproductive disorders and being teratogen

(Europaparlamentets och rådets förordning (EG) nr 1907/2006, 2006). With the implementation of REACH, there is a need to find suitable substitutes for LA-based primary explosives, i.e. “green” primary explosives.

Toxicity

The toxic effect of a compound and other physiological responses are

determined by the dose and pathway of exposure (Klaassen & Watkins, 2015). Understanding of the toxicity pathways and mechanisms is hence crucial for determination and evaluation of a products environmental impact. Within the chemical group of explosives, heavy metals are a common cation e.g. lead (Pb), silver (Ag), mercury (Hg), zirconium (Zr) and others (Ilyushin & Shugalei, 2019). Although they have desired properties in the production of explosives they are in general toxic, both for humans and for the environment (Tchounwou, Yedjou, Patlolla, & Sutton, 2012). Some examples of toxicity and usage ion explosives are given below.

Lead

Lead has been utilized in applications dating as far back as in ancient Egypt for glazing of pottery, in ancient Rome as a sweetener of food (as lead acetate) and material for piping (Greenwood & Earnshaw, 1994). However, today it is well-known that lead has several toxic properties both in the environment and in humans where it affects e.g. kidneys, liver, central nervous system, reproductive system as well as being teratogenic (Domingo, 1995; Tchounwou, Yedjou, Patlolla, & Sutton, 2012; Holecy & Mousavi, 2012).

Silver

Silver has, similar to lead, been used throughout history for different applications (Greenwood & Earnshaw, 1994). Silver has toxic properties mostly in the form as silver nanoparticles or silver ions which is cytotoxic (Sweeney, o.a., 2016;

Sokołowska, o.a., 2017). Silver ions however are mostly found in solutions, whereas the silver nanoparticles consist of elemental silver. This is important with regards to the previously mentioned explosive properties where SA forms elemental silver.

Potassium 4,6-Dinitrobenzofuroxan

Potassium 4,6-Dinitrobenzofuroxan (KDNBF) is a promising “green” and

environmental friendly alternative to heavy metal containing primary explosives, based on that it utilizes potassium as the cation instead of a heavy metal

(Matyáš & Pachman, 2013; Ilyushin & Shugalei, 2019). Because of the novelty of the compound there is a severe lack of toxicity data.

Instrumentation (settings and performance)

Gas Chromatograph (GC)

Analysis of explosives can be performed on GC-systems using a polar column (Calderanra, Gardebas, & Martinez, 2003). The ZB-50 column is suitable for analyzing polar compounds and has been used for the analysis of caffeine (Barnett, o.a., 2014). Despite having very different properties caffeine share several common chemical similarities with KDNBF, hence should the ZB-50 column be a suitable choice for GC-analysis of KDNBF. The molecular weight of the two is 282 and 194 Da for KDNBF and Caffeine, respectively. Both of them have a benzene ring, and a furoxan ring bound together. Further, they both have electronegative oxygen atoms on ketone configurations from the benzene ring structure. Both compounds also have nitrogen atoms bonded to the furoxan part, although caffeine has both of the nitrogen atoms in the benzene ring structure whereas the KDNBF has the nitrogens in nitro groups attached to the benzene ring.

Liquid chromatograph (LC)

Using an LC-system in combination with a porous graphitic carbon (PGC) (Hypercarb ®) column, has showed qualitative and quantitative results for organic explosives such as nitroaromatics (Holmgren, Carlsson, Goeda, & Crescenzi, 2005). A PGC-column retains small polar molecules by interactions on the graphitic surfaces via π-interactions and lone-pair electrons in the

molecule thus forming a strong bond (Holmgren, Carlsson, Goeda, & Crescenzi, 2005). This is the major difference of functionality from the conventional reversed phase column such as C18 silica column.

Titration

There are different ways to interpret the titration method used in the article by Lur’e, Sinditskii, and Smirnov. One of them suggests that KDNBF is first dissolved in excess 0.2 mol L-1 HCl, and then titrated with 0.1 mol L-1 NaOH.

Another possibility is that KDNBF is dissolved in water and then titrated with HCl 0.2 mol L-1 and then back-titrated with 0.1 mol L-1 NaOH. Independent of the

previous possibilities, was the results interpreted so that the first inflection point corresponds to all HCl being consumed, and the second one to the amount of DNBF in the sample (Lur’e, Sinditskii, & Smirnov, 2003).

Since the sample is dissolved in an excess of acid, the explanation would be that the H3O+ participates in an ion exchange with the K+ in the KDNBF. A likely

interpretations that there is a small difference between the nominal amount and analytical amount of H3O+ titrated. The difference between these values is

equivalent to the amount of K+ in the KDNBF. This would then be represented by

the first inflection point, where all the excess acid has been neutralized with the base. Furthermore, if the titration with OH- continues there would be a reaction

with the H+ in the newly formed H-DNBF, resulting in H2O and DNBF. This would

be indicated by the second inflection point of the titration function. Under optimal conditions, the volume to the 2nd inflection point should be two times the volume

to the 1st inflection point, if the ratio between DNBF and K+ is 1:1 (Lur’e,

Aim

The aim of the project is to develop a method for the determination of KDNBF utilizing the DNBF-anion by potentiometric quantification. Furthermore is the scope of the project to validate the results by analysis of the potassium content using atomic absorption spectroscopy (AAS). Secondly is the aim to develop and use “high performance liquid chromatograph – ultraviolet-visible spectroscopy” (HPLC-UV/VIS) and “gas chromatography – flame ionization detector” (GC-FID) methods to verify the result from the potentiometric titration.

Materials and Methods

Materials

Chemicals

All water used in the experiments was distilled to avoid unwanted side reactions and analytical errors. Buffer solutions for calibration of the pH-meter were of technical grade from Mettler Toledo, pH 4.01 (lot. no. 1E256A); pH 7.00 (lot. no. 1E256A) and pH 9.21 (lot. no. 1E248E). For titration was fuming (37 %) HCl used as well as sodium hydroxide prepared from pellets. All chemicals were of high purity and in general of analytical grade as stated below.

- Dimethyl sulfoxide (DMSO), analytical grade, VWR Chemicals - Toluene, EMPARTA® ACS grade, Merck

- Acetonitrile, LiChrosolv® Reag. Ph Eur grade, Merck

- Methanol, LiChrosolv® Reag. Ph Eur grade, Merck

- Caffeine, Merck

- HCl (≥ 37.0 %), Reag. Ph Eur EMSURE® grade, Merck

- HNO3 (65 %), Reag. Ph Eur EMSURE® grade, Merck

- NaOH (≥ 99.0 %), EMPLURA® grade, Merck

- Cesium nitrate (CsNO3), > 99 %, Aldrich Chemistry Labware

During the experiments, polypropylene (PP) lab-ware was used to minimize interactions especially between cations and charged surfaces e.g. glass. Test tubes made of PP (50 mL Sarstedt ®), syringes (WVR-line) and 0.2 µm PP-filters (WVR-line) were used. Glass vials (2 mL) and PTEF/silicone screw-caps (VWR-line) were used for gas and liquid chromatography due to the lack of equivalent equipment made of polypropylene. Stock solutions of NaOH were kept in glass, the rest of stock solutions were kept in 50 mL test tubes (PP).

Methods

Titration

Titration was performed in 50 mL Sarstedt ®-tubes, where KDNBF was dissolved in HCl and titrated with NaOH. The concentration of the acid was determined by titration with NaOH which concentration was set by weight to weight dilution of pellets in distilled water. The sample for titration was prepared by adding approximately 3 mg KDNBF to a test tube and then dissolve the salt by adding 0.1 mol L-1 HCl. From this stock solution was then a subsample of approximately

4 g withdrawn and titrated. All masses were corrected for density.

The volumes needed to reach the inflection points were obtained by evaluating the data using the Gran-plot. The Gran-plot uses the cumulative amount of added titrant in relation to the total volume that has been weighted against the corresponding pH (see equation 3).

A test tube containing the subsample previously described, was placed on the scale operating at 0.01 – 41 g (±0.01 mg). Then the pH electrode was lowered into the solution and measured. Once stable, the pH-electrode was raised and then the scale was tared, before the next addition of NaOH. This was repeated thorough the whole titration. Additions of NaOH was done by using a pipette.

𝑌 = (𝑉𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑖𝑛 𝑠𝑎𝑚𝑝𝑙𝑒+ 𝑉𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑎𝑑𝑑𝑒𝑑 𝑏𝑎𝑠𝑒)10−𝑝𝐻 (Eq. 3) Metal analysis

Samples for metal analysis were prepared by dissolving KDNBF in a mixture of 10% HNO3 and 1.25 g L-1 CsNO3 (ionization buffer), dissolved in deionized

water. The samples were filtered through 0.2 µm filter and sent for analysis at Eurenco Bofors AB.

GC-FID

The samples were prepare by dissolving KDNBF in DMSO. The samples from batch A and B, which were previously bought, were dissolved in DMSO, 14.99 and 15.02 mg respectively in 0.05 L, resulting in concentrations of 299.8 and 300.4 mg L-1. After vigorously shaking, the samples were filtered through 0.2 µm

filter and 1 mL of the solution transferred to the vial. As internal standard (IS) was caffeine used and 100 µL of a solution prepared as follows was added to each sample after filtration, but prior to analysis. The IS was prepared by dissolving 8.76 mg caffeine in 52.84 g DMSO, resulting in a concentration of 182.36 mg L-1. Data valuation is done by ratio calculations between area and

concentration of the IS.

The samples was prepare by dissolving KDNBF in different solvents i.e. acetonitrile, water and the mobile phases (described elsewhere (Holmgren, Carlsson, Goeda, & Crescenzi, 2005). Minor changes were made compared to the reference mobile phase as follows: A from: 49.5 % H2O, 9.9 % methanol,

39.6 % acetonitrile, 1 % dichloromethane to: 50 % H2O, 10 % methanol, 40 %

acetonitrile. Mobile phase B was used as in the reference i.e. 73 % methanol, 25 % acetonitrile, 2 % toluene; and C: 25 % acetonitrile, 75 % toluene. Similar to the samples for GC-analysis were the samples filtered through 0.2 µm filter. IS was prepared by dissolving caffeine in the same solvents as was used for the

samples. Then 100 µL of prepared IS was added to 1 mL of sample prior to analysis.

Instrument

Protection

All laboratory work was conducted in a climate with relative humidity of 50-60 %, and temperature at 22 oC, with these conditions there is decreased risk of

electrostatically spontaneous discharges when weighting and working with primary explosives. Furthermore, ESD-protective clothing, and ESD-armlet was worn at all-time throughout laboratory work.

Scale

The scale was a Mettler Toledo XSE105 Dualrange, with max weight of either 0.01-41 or 0.1-120 g, with a precision of ± 0.01 and 0.1 mg respectively. When weighting the sample, the lower limit was used for increased precision.

pH – Meter

The pH-meter used for titration was a SevenCompact pH meter s220 by Mettler Toledo. Calibration was performed before use as a three-point calibration, (4.01, 7.00, 9.21 buffers) with the requirement of slope >95 %.

AAS

The metal content of the samples was analyzed on a PinAAcle 900F-AAS from Perkin Elmer, operated in flame mode, with a hollow cathode lamp at λ=766.5 and 589.0 nm for potassium and sodium, respectively. With injection speed of 8 mL min-1 and injection/integration time of 3 s.

GC-FID

A Nexis GC-FID-2030 Gas Chromatograph from Shimadzu was used with the following settings: 1.0 µL injection volume, 3 and 5 rinses pre-, and post-run respectively, and 8 µL washing volume. Split mode was used with a ratio of 9.9, and 195 oC injection temperature. The carrier gas consisted of helium with a total

flow of 19.9 mL min-1 and pressure of 65.8 kPa. The column flow was 1.55 mL

min-1 while the purge flow was 3.0 mL min-1. The following oven temperature

program was applied: initial temperature at 100°C with an increase of 7.5 °C min -1 to 280 °C and hold for 10 min. The column used was a ZB-50 from

with column outlet temperature at 310.0 °C, sampling rate at 40 ms, the makeup gas was N2 with a flow at 30.0 mL min-1, and H2 flow at 47 mL min-1, and air flow

at 400.0 mL min-1. Integration was performed with Chromatopac software, with

peak width of 2.5 sec, and slope of 500 µV min-1. With a drift at 0 µV min-1 the

minimal area to height ratio is set to 1000 counts, and the detection limit is calculated by area. The identification was set to 0.1 minutes broad peaks.

HPLC-UV/Vis

A Shimadzu HPLC-UV/VIS system was used with the following pars; system controller CBM-20Alite, pump LC-20AD, auto sampler SIL-20A, oven CTO-20AC, detector SPD-20A. The mobile phases used were the same as mentioned in method section (Holmgren, Carlsson, Goeda, & Crescenzi, 2005). The gradient program that was used was the fast program described elsewhere, with the following protocol; 0.00 - 3.20 min, 32 % A & 68 % B: 3.20 - 5.20 min gradient to 22 % A & 78 % B: 5.20 - 6.00 min held at 22 % A & 78 % B: 6.00 - 8.00 min gradient to 0 % A & 40 % B & 60 % C: 8.00 - 8.40 min gradient to 100% C: 8.40 - 12.00 min held at 100 % C: 12.00 - 12.20 min gradient to 32 % A & 68 % B: 12.20 - 17.00 min held at 32 % A & 68 % B, with a flowrate of 0.800 mL min-1

(Holmgren, Carlsson, Goeda, & Crescenzi, 2005) as seen in Figure 1. Rinse flow was set to 35 µL s-1 and sampling flow to 5 µL s-1. Purge time was 10.0 min,

rinse dip time 15 s, oven temperature was set at 25 °C, the detector wavelength was set at 275 nm, the detector was set at positive polarity, using a D2-lamp, with 1.0 sec response time, with an auxiliary range of 1.0 AU V-1, and a recorder

range of 1.0000, whilst having a ratio range of 10, and a ratio threshold of 0.0001 AU. Integration was performed with Chromatopac software, with a width of 5 sec, and a slope at 200 µV min-1. A drift of 0 µV min-1, minimal area to height ratio is

set at 1000 counts, and calculated via area. The peaks are integrated using window mode, and 5.00% window, with closest peak as selection mode. The used column was a porous graphitic carbon (PGC), Hypercarb™ 100 × 4.6 mm, with a particle size of 3 µm.

Figure 1. Mobile phase composition during the HPLC-UV/VIS program. 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 % o f re s p e ci tv e m o o b ile p h a s e Time (min)

Mobile phase Composion

A B C

Results and discussion

Observations

Because of the novelty of the compound, no solubility data is available. Further, the solubility of KDNBF is found to be very low in water, toluene, and

dichloromethane. The low solubility makes the development of a quality control protocol difficult, since most of the methods used for similar substances involve the previously mentioned solvents. However, there are possible alternatives, and one might be dimethyl sulfoxide (DMSO), with respect to solubility. Concerning the titration method, this implements the suitability of dissolving KDNBF in acid rather than in water, as described in the introduction. By doing that, it would also mitigate the dangers of working with amounts greater than 20 mg of KDNBF.

It has been suspected that although formamide can be used to dissolve KDNBF it also induces decomposition, as evidenced by previous in-house studies. The Karl-Fisher titration (for water content analysis), clearly indicated that

decomposition took place. The degradation of KDNBF in formamide was also observed when testing this solvent for GC-FID analysis. After dissolving KDNBF in formamide and analyzing the samples within a day qualitative data was

obtained that indicates the presence of KDNBF at the expected retention time. However, when reanalyzing the samples after 17 days no traces of KDNBF was found. This might be explained by the fact that formamide has a similar chemical structure as ketones that are known to degrade TNT, which is an explosive similar to KDNBF (Holmgren, Carlsson, Goeda, & Crescenzi, 2005).

Titration

Since only a maximum 20 mg KDNBF can be handled with safety, titration is not easily conducted, although it is a desirable method. Further, this low amount of analyte makes it necessary to use concentrations of HCl and NaOH well below the concentrations reported by Lur’e, Sinditskii, & Smirnov, in order to correctly identify the inflexion points. Using a manual pH-meter the titration was to

performed using 0.1 mol L-1 HCl and 0.04 mol L-1 NaOH. As seen in Figure 2 the

use of these concentrations resulted in a titration curve showing two inflection points. Increased resolution could be obtained by using 0.0025 mol L-1 NaOH, as

Figure 2. Titration of 7.12 g solution containing Potassium 4,6-dinitrobenzofuroxan at a concentration of 45.2 mg L-1 in 0.1 mol L-1 HCl, with 0.04 mol L-1 NaOH.

During the titration with 0.0025 mol L-1 NaOH, four inflection points were

identified, see Figure 3. This finding is in contrast to the previous theory, where only two inflection points were expected. The first one corresponding to K+, and

the second to H3O+. These results suggest that the initial ion exchange process

is not only between K+ and H+. Evidently there is also an electrostatic attraction

of H3O+ to the nitro-group, the single oxygen in the furoxan part of the molecule,

as visualized in Figure 4.

The titration was performed three times, using 0.04 mol L-1 NaOH once and

0.0025 mol L-1 NaOH for the other. When calculating the ratio between the

weighted amount of KDNBF (i.e. nominal concentration) and the measured concentration (calculated from the titration using a Gran-plot), the following ratios were obtained; 0.880.04 M; 0.840.0025 M; and 1.080.0025 M. The average ratio is 0.93

with an RSD of about 14 %. This means that titration, is slightly more prone to underestimate the K+ content with respect to the weighted amount. Although

there are several reasons for this, the most rational explanation behind the observation would be the presence of possible impurities in the KDNBF.

0 2 4 6 8 10 12 11 12 13 14 15 16 17 18 19 20 21 pH

Cumulative amout (g) of 0.04 mol L-1_NaOH Titration of KDNBF with NaOH

Figure 3. Titration curve of 45.2 mg L-1 KDNBF dissolved in 0.1 mol L-1 HCl, titrated with 0.0025 mol L-1 NaOH. An example of the Gran-plot is included for visualization.

Figure 4. Proposed ion exchange during dissolution of KDNBF in acid. Metal analysis

To manufacture KDNBF, NaDNBF is reacted with K+ forming KDNBF and Na+.

However, there are reasons to suspect that traces of NaDNBF will always be present in the KDNBF because of the pathway of synthesis. By analyzing the metal content in dissolved KDNBF it would be possible to have a quality check, assuming there is a 1:1 ratio between the sum of the metal and DNBF ions. The nominal mass to mass ratio of K+ in KDNBF is calculated to be 13 %. The

analytical K+ content was lower than the nominal, with a ratio of 1:0.937. To

further investigate what the remaining 7% consisted of, analysis of other

elements such as Na+ is necessary. Analysis of Na+ and K+ in the same sample 0 2E-08 4E-08 6E-08 8E-08 0.0000001 1.2E-07 1.4E-07 0 2 4 6 8 10 12 13 13.5 14 14.5 15 pH

Volume (mL) added 0.0025 M NaOH

Titration of KDNBF Solution with example Gran-plott

KDNBF:HDNBF:DNBF Gran-Plott

manages to explain 97 % of the expected sum of these elements. Hence, it is proposed that analysis of both Na+ and K+ would be beneficial in order to better

quantify the DNBF when using the metal content. The total mass to mass ratio with respect to both Na+ and K+ is 12.4%. The percentage of Na+ in the sum of

Na+ + K+ is calculated to be 3.20 %. With this method on its own however, there

is a risk of underestimate the purity of the DNBF, based on the performance and sensitivity of the used instrument.

It is possible to use the titration method to estimate the metal content in KDNBF. Using the titration as described earlier and combined with metal analysis data, the ratio between the nominal metal content and the analytical metal content can be calculated. Assuming the presence of only K+ the ratio would be 1.047.

However, by using the sum of Na+ and K+, the ratio equals 1.014. Hence, this

approach is more appropriate to use for the calculation of the metal content.

GC-FID

There is no certified reference material available for KDNBF up to this day. Hence, an in-house manufactured batch that had been qualified with H1-NMR at

Cambrex Karlskoga AB, was used for preparation of calibration solutions. A four point calibration curve with a R2 > 0.9968 using caffeine as an internal standard

was achieved (see Figure 5). Further, two other batches, A and B, that were bought from two other manufacturers prior to this project were analyzed using the previously mentioned GC-program and the resulting calibration function was used for quantification.

For this method the limit of quantification is somewhere between 87.4 mg L-1 and

198.8 mg L-1.The large span iscaused by the absence of signals for the

concentrations of 87.4 mg L-1 and 198.8 mg L-1, respectively. This is most likely

related to the small amount that can be handled. Both of the samples A and B gave responses above the limit of quantification (see Figure 6 and Figure 7). The results showed that batch A and B had concentrations of KDNBF of 255.2 mg L-1

and 197.6 mg L-1, respectively. The difference from the nominal concentrations

of 299.4 and 300.4 mg L-1, respectively, is most likely explained by ion-exchange

and/or decomposition. Both batches were transported to SAAB AB Dynamics during an unknown time, for batch B as an aqueous solution (Norberg, 2020). Depending on the composition of the water there are possibilities that some of the KDNBF could have changed into the water soluble NaDNBF or HDNBF by ion exchange with either sodium or hydrogen ions from the water.

Figure 5. Calibration curve from GC-FID with R2 > 0.9968, and caffeine as internal standard.

Figure 6. GC-chromatogram of KDNBF in DMSO from batch A, showing caffeine with retention time at 20.731 and DNBF at 23.401 min. The concentration of DNBF in batch A is calculated to 255.2 mg L-1.

Figure 7. GC-chromatogram of KDNBF in DMSO from batch B, showing caffeine with retention time at 20.731 and DNBF at 23.404 min. The concentration of DNBF in batch B is calculated to 197.6 mg L-1.

HPLC-UV/VIS

It turns out that KDNBF is not suitable for analysis with HPLC, even though it has some shared properties with members of the nitroaromatic explosives family. These compounds show good separation and quantification when using the same instrumental settings and mobile phases as in this study (Holmgren, Carlsson, Goeda, & Crescenzi, 2005). It can be assumed that the poor solubility of KDNBF in toluene, and that acetonitrile most likely enhances degradation of the KDNBF, resulted in non-detectable signals from KDNBF during HPLC-analysis.

It was found that the signal for caffeine was present with the same retention time for each trial although no signal for KDNBF was to be found. Since the column of choice needs non-polar solvents in the mobile phase to elute this kind of

compounds, it can be of interest for further investigation – to use a non-polar solvent that still can dissolve KDNBF.

Conclusions

The potentiometric method gave an accuracy range between 0.84 and 1.06 for the ratios of detectable amount and nominal, but in combination with metal analysis an accuracy of 1.01 could be calculated.

The developed method for GC-FID had a limit of quantification in the range of 87.4 and 198.8 mg L-1, and gave a linear calibration curve (R2 > 0.9968) using

caffeine as the internal standard.

By combining gas-chromatography, potentiometric titration and metal analysis is it therefore possible to obtain both quantitative and qualitative analytical data for the DNBF-anion and hence extract maximum amount of information regarding e.g. the purity of KDNBF.

Analysis of KDNBF using HPLC-UV/VIS suffers from a serious drawback related to detection of the analytical signal related to solvent changes, and no detection of the analyte, even at the concentration of 400 mg L-1.

Acknowledgements

I would like to thank the personal at SAAB AB that have coached me during my work. Further on I would like to thank my supervisor at university for being supportive through the whole work. I would like to thank Chematur Engineering AB and Eurenco Bofors AB for being helpful with analyses. I would like to thank Erik Holmgren at FOI for advices for HPLC instrumentation, and I would also like to thank Samantha Sambois and Daniel Baker at Franklin County Coroner for advices for GC instrumentation.

References

2.10.1 Properties and Behavior of Explosives. (2020, 02 12). Retrieved from

FRTR Remediation Technologies Screening Matrix and Reference Guide, Version 4.0: https://frtr.gov/matrix2/top_page.html

Agrawal, J. P. (2011). High Energy Materials; Propellants, Explosives and

Pyrotechnics (1 ed.). New Delhi,, India: Wiley-VCH.

Barnett, R. Y., Baker, D. D., Kelly, N. E., McGuire, C. E., Fassette, T. C., & Gorniak, J. M. (2014, October). A Fatal Intoxication of 2,5-Dimethoxy-4-Chloroamphetamine: A Case Report. Journal of Analytical Toxicology,

38(8), 589-591.

Bennett, H. (1939). The Chemical Formulary (1 ed., Vol. 1). New York: Chemical Publishing Company, INC.

Calderanra, S., Gardebas, D., & Martinez, F. (2003, October). Solid phase micro extraction coupled with on-column GC/ECD for the post-blast analysis of organic explosives. Forensic Science International, 137(1), 6-12.

Domingo, J. L. (1995). Prevention by chelating agents of metal-induced developmental tocicity. Reproductive Toxicology, 9(2), 105-113. Europaparlamentets och rådets förordning (EG) nr 1907/2006 av den 18

december 2006, 1907/2006 (European Parlament 12 18, 2006). Retrieved 02 06, 2020

European Parlament and the council of the european union. (2013, 12 20). REGULATION (EU) No 1293/2013 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL OF THE EUROPEAN UNION of 11 December 2013 on the establishment of a Programme for the Environment and Climate Action (LIFE) and repealing Regulation (EC) No 614/2007. Official

Journal of the European Union, 185-208.

Greenwood, N. N., & Earnshaw, A. (1994). Chemistry of the Elements. Tarrytown, New York: Pergamon Press.

Holecy, E. G., & Mousavi, A. (2012). Lead Sources, Toxicity, and Human Risk in Children of Developing Countries: A Mini-Review. Environmental

Forensics, 13(4), 289-292.

Holmgren, E., Carlsson, H., Goeda, P., & Crescenzi, C. (2005). Determination and characterization of organic explosives using porous. Journal of

Chromatograpgy A, 1099, 128-135.

Ilyushin, M. A., & Shugalei, I. V. (2019). Eco-Friendly Energetic Substances for Initiation Devices. Handbook of Ecomaterials(138), 3433-3448.

Klaassen, C. D., & Watkins, J. B. (2015). Casarett & Doull's Essentials of

Toxicology, Third Edition. United States of America: McGraw-Hill

Education.

Larsson, U. (2010). Alfred Nobel: Nätverk och Innovationer. Norrköping: Nobelmuseet.

Lur’e, B. A., Sinditskii, V. P., & Smirnov, S. P. (2003). Thermal Decomposition of 2,4-Dinitrobenzofuroxan and Some of Its Compounds with Metal

Hydroxides. Combustion, Explosion, and Shock Waves, 39(5), 534-543. Matyáš, R., & Pachman, J. (2013). Primary Explosives. Pardubice: Springer. Mcfee, J. M. (2010). Shock Wave Science and Technology Reference Library

Mehilal, A. S., Suman, P., & Nirmala, S. (2002). Synthesis, characterisation, thermal and explosive properties of 4,6-dinitrobenzofuroxan salts. Journal

of Hazardous Materials, 221-227.

Norberg, A. (2020, 01 28). KEX-Demonstation. (A. Norberg, Performer) Saab Dynamics, Karlskoga, Örebro County, Sweden.

Sejlitz, I. (2014). Explosivämneskunskap (3th ed.). Karlstad: Strokirk - Knappen. Sokołowska, P., Białkowska, K., Siatkowska, M., Rosowski, M., Kucińska, M.,

Komorowski, P., . . . Walkowiak, B. (2017). Human brain endothelial barrier cells are distinctly less vulnerable to silver nanoparticles toxicity than human blood vessel cells A cell-specific mechanism of the brain barrier? Nanomedicine: Nanotechnology, Biology, and Medicine, 13, 2127-2130.

Sweeney, S., Leo, B. F., Chen, S., Abraham-Thomas, N., Thorley, A. J., Gow, A., . . . Tetley, T. D. (2016). Pulmonary surfactant mitigates silver nanoparticle toxicity in human alveolar type-I-like epithelial cells. Colloids and Surfaces

B: Biointerfaces, 167-175.

Tchounwou, P. B., Yedjou, C. G., Patlolla, A. K., & Sutton, D. J. (2012). Heavy Metals Toxicity and the Environment. National Institutes of Health, 101, 133-164.