Low-perchlorate blue-flame pyrotechnic compositions

Abbreviations

AGN

aminoguanidinium nitrate

AP

ammonium perchlorate

KP

potassium perchlorate

GN

guanidinium nitrate

BCC

basic copper carbonate (malachite)

Dex

dextrin

PVC

polyvinylchloride

Ω _(CO2)

oxygen balance (to CO₂)

L _sp

specific luminous intensity

I _v

light intensity

p _e

color purity

1 Introduction

Blue-flame pyrotechnic compositions are somewhat different from red, orange, yellow, green, and white-colored flame compositions, in part due to a limited temperature window at which blue color emitters (CuCl) are stable, in part due to different properties of Cu and its compounds in comparison to alkali [1] and alkaline earth metals [2] that are used as emitters in pyrotechnics such as Na, Ca, Sr and Ba [3], [4], [5], [6], [7], [8], [9], [10]. This restrains compositions in a way that practical blue-flame compositions use chlorates (KClO₃) or perchlorates (NH₄ClO₄, KClO₄) as oxidizers that contain from 38 to 70% [11], [12], [13] of the composition. KBrO₃ is less desired in blue-flame compositions due to its high sensitivity to friction, questionable chemical stability, higher price, and toxicity [5, 14].

Being chemically stable, oxygen-rich and well proven over time, chlorates and perchlorates have earned their places in almost all blue-flame pyrotechnic compositions [13, 15, 16]. Only very few, novel compositions have been reported that are chlorate- and perchlorate-free [5, 14, 17], [18], [19]. While the chromaticity parameters (x/y), p_e for the dominant wavelength have been reported, there is no data on the light intensity of these novel compositions.

Perchlorate-free compositions, if composed, might have an advantage of possessing flame’s higher color purity due to absence of potassium that has several weak emissions in the visible spectrum at 404, 580, and 694 nm [20], together with the doublet in the NIR region at 767 nm [20]. Moreover, perchlorates can be easily avoided in red and green-flame compositions with strontium and barium nitrates as proven in the past [3, 6, 18, 21], whereas in case of blue flames, copper nitrate (basic) is more difficult.

Perchlorate-free blue-flame compositions with GN and AGN are less energetic, therefore different sample preparation methods were suggested. Instead of pellets, it was chosen to burn loose powdered compositions in different amounts. In this way, the relative number of pores is significantly increased and burning propagates more rapidly like commercial black powder. However, the tests are less reliable and do not provide coherent measurement data, as the burning becomes more chaotic. The theoretical maximum density (TMD), the linear burning rate and L_sp cannot be estimated. Nonetheless, loose powders for a flash or a colored flame of short duration are used in practice.

Another approach is to press the composition in a different shape, preferably with a cavity. A cavity helps to conserve the heat given off by a chemical reaction for pre-heating the unreacted layer of the composition. This becomes very useful when testing such compositions that are hard to ignite or that do not sustain combustion. The disadvantage is that such test units are more difficult and time-consuming to produce.

Pellets have become standardized test samples that are used to compare compositions. In fact, from a practical perspective, pressed pellets (a.k.a. stars in firework terms) are most popular pyrotechnic components used in fireworks. The test also provides an idea of how practical the composition is. For this reason, most tests for this experiment were performed using pressed pellets. Testing loose powders gave some rough guidance for further investigations.

An experiment was suggested where AP compositions are gradually replaced by aminoguanidinium nitrate (AGN)-based compositions. Guanidinium nitrate was also a good candidate for this experiment. However, AGN was suggested for its more pronounced reactivity. Such experiments provide information on how the amount of AP affects the I_v, and L_sp values in such systems. Furthermore, a copper-AGN (CuAG) compound was synthesized as a potential energetic material for perchlorate-free blue-flame compositions.

2 Results and discussion

Two compositions were selected for this experiment. One simple AP-rich composition (A) that provides a stable combustion and high-saturation blue flame of p_e = 50%, L_sp = 100 cd s g⁻¹. Second, a perchlorate-free, AGN based composition (B) that burns quite slowly producing a dim blue flame.

In-between compositions were prepared by mixing compositions A and B in different weight ratios. Compositions are presented in Table 1, and their burning characteristics are presented in Table 2.

Compositions 1–3 burned with steady tall flames when the AP content was >60%. Composition 9 with no AP burned with a steady flame that looked like a burning tablet of hexamine. We assume that the combustion is based on slow decomposition of AGN catalyzed by Cu which generates flammable gases. We also observed experimentally that a small percentage of shellac in the compositions made the gases easier to ignite. This can be attributed to the good fuel properties of shellac as was described by Shimizu [22]. He states that a remarkable property of shellac is that the ratio of hydrogen to carbon is one of the largest among the common pyrotechnics fuels, that can help to achieve less sooty high-temperature flames.

Pellets (compositions 4–8) did not sustain combustion on their own. For that reason, in order to sustain combustion, a heated metal wire was used to catalyze combustion on the pellet’s surface (Figure 1). It did not disturb the blue flame color as the metal wire was never hot enough to affect the emissions observed in the flame. And even with the support of the metal wire, some pellets burned in an oscillatory fashion as observed in Figure 2.

Figure 1:

Steel wire assists combustion of compositions 4–8. The wire is lifted, pre-heated with a butane torch and then released. A weight (10 g) on the right forces the glowing wire by gravity to stay in contact with the burning surface.

Figure 2:

Curves depicting light intensity at λ = 450 nm over time from compositions 1 to 9 during the burning tests.

Figures 3 and 4 show that there are significant drops in the I_v and L_sp with decreasing amounts of AP. Apparently, the stable pyrotechnic reaction is disturbed when the AP content is <60%. With an AP content in the range of 60–20%, oscillatory burning behavior is observed as was shown in our previous work [16]. However, even though the light intensity is significantly lower, the drop in L_sp is not so pronounced. The I_v is lost over the longer burning time. Nonetheless, when the AP content is lowered even further, both L_sp and I_v drop significantly.

Figure 3:

Light intensity dependence on the amount of ammonium perchlorate (AP) in the composition.

Figure 4:

L_sp dependence on the amount of ammonium perchlorate (AP) in the composition.

This experiment clarifies the role of strong oxidizers in order to achieve a blue pyrotechnic flame without use of metal fuels or nitrocellulose. In order to increase the reactivity of AGN compositions, different catalysts i.e. copper salts, oxides and metallic copper powder were tested, but the results are not included in this paper. All had a similar catalytic strength, with CuO being the top candidate. As fuels and additives, hexamine, Mg, B (amorphous), Al powders, polyvinylenedichloride (PVDC), NC and GN were also tested. Hexamine was useful in small amounts and was later replaced by shellac. A composition with 5% Mg burned very well and quite bright, however it was almost white as Mg sparks were as bright as the flame. Boron produced a greenish yellow flame. Dark aluminum powder (1–5 µm) was useful in the 1–2.5% range for increasing the temperature at the burning surface. PVDC and it’s alternative PVC performed well. NC was also satisfactory, but with more than 10% added, it reduced the flame color purity.

Finally, a CuAG complex was synthesized (experimental section) as a potential energetic material for perchlorate-free blue-flame compositions. Since the synthesis was performed in suspension (solid BCC reacted to form solid CuAG complex), it was evident that not all BCC had reacted. When CuAG was dissolved in water, a green precipitate (BCC) was observed at the bottom of the flask. Purifying was also troublesome since filtered CuAG tends to hydrolyze and change its color in solution. Since purification was troublesome and it was impossible to obtain a pure compound, it was impossible to get X-ray diffraction patterns for this compound that would identify the molecular structure.

Elemental analysis, ICP and DTA (Figure 5) investigations have been performed for this compound. Two to three samples from different synthesis batches were tested. Averaged elemental analysis arrive at N 41.7%, C 7.8%, H 3.4%. The ICP measured Cu content was 27.2%, while the theoretical value was estimated to be 18.9% for an expected product of the formula [Cu(AG)₂](NO₃)₂ with M_r of 335.7 g mol⁻¹. The DTA curves are presented in Figure 3. Exothermic decomposition was observed to start at 130 °C and another, more rapid decomposition step took place at 260 °C. As compounds with T_dec = 200 °C are acceptable for practical use, the CuAG complex appears to be too unstable to meet this requirement.

Figure 5:

DTA curves of copper-aminoguanidinium (CuAG) (bold) and aminoguanidinium nitrate–basic copper carbonate (AGN–BCC) in the ratio of 4:1 (dotted line). CuAG shows an exotherm as early as T = 120 °C, while the mixture of AGN–BCC follows around T = 170 °C.

Several compositions containing CuAG were tested. Low decomposition temperatures render the compositions to be more reactive than AGN–BCC compositions. Several CuAG compositions are presented in Table 3 and their burn tests are depicted in Figure 6 together with those of compositions 1 and 9.

Figure 6:

Combustion of compositions: A) 1; B) 9; C) 11; D1–D3) 14.

Among the compositions 10–14, 11 was quite good as it had the capacity to sustain a stable tall blue flame.

Also, composition 14 was of special interest. When it was heated with a butane torch it did not ignite immediately. It melted and bubbled on the surface and then after ∼5 s of heating a rapid decomposition took place, giving off a large green flame for a short time (<0.5 s). As the composition was chlorine-free, the green color was a result of the formation of CuOH. The flame had the same red tip as flames with AP, due to oxidation to CuO, which gives a weak red-orange emission. This experiment proves once again that CuOH, in the right conditions, can be used as a green flame emitter [5, 14, 23].

3 Conclusions

An experiment to identify the necessity of perchlorate in a blue-flame composition was performed. AP-based compositions were gradually diluted by AGN-based blue-flame compositions and their burning curves, light intensity and burning behavior were recorded.

60–80 wt% AP was found to be crucial in order to produce a bright blue flame. When the amount is reduced further, the steady combustion is disturbed and the composition burned in a strobing manner. The burning rate also became significantly lower.

Some materials were investigated as catalysts for AGN. CuAG was suggested as an energetic material, as a Cu source and as a catalyst for AGN based composition. While burning rate and flame light intensity were low, a stable blue flame was produced with an AP content as low as 10%.

A relatively bright green flash was observed when a composition containing AGN/CuAG was heated. The green color was observed due to the emission of CuOH species.

4 Experimental section

CAUTION! The mixtures described herein are potential explosives, which are sensitive to environmental stimuli, such as impact, friction, heat, and electrostatic discharge. Although we encountered no problems in the handling of these materials, appropriate precautions and proper protective measures (safety glasses, face shields, leather coats, Kevlar gloves, and ear protectors) should be taken when preparing and manipulating them.

AP was synthesized by neutralizing HClO₄ (60%) with ammonia solution (25%) in a molar ratio of 1:1 in 10 g batches; yield 98%. AGN was synthesized by reacting 3.54 g of aminoguanidinium bicarbonate (moistened with 0.5 mL of water) with 2 mL nitric acid (60%). The final product was dried in a desiccator before use. The aqueous solution of AGN had pH 7–8; yield 98%. Basic copper carbonate and PVC powder were from Sigma Aldrich. Shellac, −120 mesh, powder was technical grade. All chemicals used were ground with a mortar and pestle and passed through a 30 mesh screen before conducting experiments.

Composition A (12 g) after mixing was moistened with 0.5 mL of distilled water and hand-mixed with a spatula until the mass became homogeneous. Then it was dried at room temperature for two days. Composition B was mixed on the day of the experiment in order to avoid reaction of AGN with BCC [16]. After mixing the dry components, 2 mL of MEK was added for 12 g of composition and kneaded. After drying at room temperature for 1–2 h (RH 40–60%), the intermediate compositions were prepared by gently mixing each of them in a mortar to break any lumps that had formed during the binding process. The compositions 1–9 were pressed. Pellets of 2 g (13 mm in diameter, ∼10 mm in height, ρ ∼ 1.6 g cm⁻³) were pressed in one increment by a consolidation dead load of 2 tons.

CuAG synthesis: 3 g of acid-free AGN was dissolved in distilled water and 1 g of finely ground BCC was added to the solution while being actively stirred. The color of undissolved BCC changed first to light violet and then to purple. Bubbling was observed. After stirring the suspension overnight, the reaction was finished. The purple precipitate was filtered off. Yield: ∼2.8 g.

Spectrometric measurements were carried out using a HR2000+ES spectrometer with an ILX511B linear silicon CCD-array detector controlled by software from OCEAN OPTICS. The integration time for recording the emission spectra was set to 50 ms. The detector-sample distance was 1 m. The DTA analysis was done with a 552-Ex differential thermal analyzer from OZM at heating rates of 5 °C min⁻¹.