Introduction

Fireworks makers fill the night sky with myriad effects in displays that are popular all over the world. Although the art dates back to ancient China, most of the effects you'll see in a typical display are inventions of this century. A typical example is the development of coloured flames. Before the 19th century, only various yellows and oranges could be produced with steel and charcoal. Chlorates, an invention of the late 18th century and an industrial product of the 19th century, added basic reds and greens to the pyrotechnist's repertoire. Good blues and purples were not developed until this century, although it is not unusual to find unsafe display formulas for blue stars in earlier literature.

Basic principles of pyrotechnic light production

The light emitters can be grouped into two main categories: solid state emitters (black body radiation) and gas phase emitters (molecules and atoms).

Black body radiation and the grey body concept

A black body is an ideal emitter which is capable of absorbing and emitting all frequencies of radiation uniformly. The excitance (M) of the black body, the power emitted per unit area, is defined as

M = sT4                                                         (1)
where s is the Stefan-Boltzmann constant and T is the temperature. Thus, we could obtain a twofold increase in radiation by merely increasing the flame temperature from, say 2000 K to 2400 K. Furthermore, the radiation also shifts from infrared to visible light as the temperature increases. The calculated emission spectrum (the energy per unit volume per unit wavelength range) has the following shape:

Fig. 1. Black-body radiation.

In the real world, simplified models are not of much help. Many solids do emit light in the same relative proportions as a black body, but not in the same amounts. The emissivity of a solid substance is the factor relating observed and theoretical radiant energy. The emissivities of many refractory metals and metal oxides are higher in the short wavelength end of the visible spectrum - that is, they look bluer than expected when heated.

Table 1 gives a summary of visual temperature phenomena of solid bodies - for instance, a glowing piece of charcoal, a good approximation to the black body.


T, K	 oC	   Subjective colour

750	 480	   faint red glow
850	 580	   dark red
1000	 730       bright red, slightly orange
1200	 930	   bright orange
1400	 1100	   pale yellowish orange
1600	 1300	   yellowish white
> 1700    > 1400     white (yellowish if seen from a distance)

Table 1. The perceived colour of heated solid bodies.

The atomic and molecular emitters

As you can easily see from Table 1 (and very probably know from experience), it is not possible to produce anything but shades of orange and yellow with grey-body emitters. (In principle, we could generate blue light with a hypothetical black or grey body at 9000 K and up, which is the temperature of blue stars, but such temperatures are unattainable for fireworkers.) For other colours, we need specific emitters of coloured light.

Surprisingly few emitters are used in pyrotechnics, given the vast range of atomic and especially molecular spectra available. In fact, the production of some colours is still a problem - next time you see a fireworks display, count all turquoises and ocean greens you saw. There are not many, because there are no commercially useful emitters available in the 490-520 nm region (blue-green to emerald green).

Table 2 summarises the sources of coloured light used in today's fireworks.



Colour          Emitters used                      Wavelength range


Yellow          Sodium D-line atomic emission 	    589 nm

Orange          CaCl, molecular bands               several bands, 591-
                                                    599 nm, 603-608 nm
                                                    being the most intense

Red	        SrCl, molecular bands	            a: 617-623 nm 
                                                    b: 627-635 nm
                                                    c: 640-646 nm

Red	        SrOH(?), molecular bands	    600-613 nm

Green	        BaCl, molecular bands	            a: 511-515 nm
                                                    b: 524-528 nm
                                                    d: 530-533 nm

Blue	        CuCl, molecular bands	            403-456 nm,
                                                    several intense
                                                    bands, less intense
                                                    bands between 460 nm
                                                    and 530 nm

Table 2. Sources of coloured light.

The chromaticity diagram and colour perception

The human eye may not be the best spectroscope invented, but it is the best instrument for designing coloured fireworks. Although a spectroscope can show the presence or absence of certain lines or bands in the flame spectrum, it cannot decide whether the colour obtained looks pleasing to the human observer. Pure, monochromatic colours a'la lasers are only a dream for pyrotechnists, but well-designed impure colours do not lag much behind.

The chromaticity diagram shown below has been designed with human colour vision system (three base colours) in mind. It is not necessary to specify the intensities of all three base colours, because the hue is not affected by the brightness of the light (the sum of all intensities). We can conveniently use the fractional intensities of two primary colours, and this gives us a chart in two dimensions. The sum of all three intensities must equal one, so the third fraction can be easily calculated.

In order to avoid negative primary colour fractions, the International Commission on Illumination published a standard chromaticity diagram in 1931 with three unreal primary colours. The above diagram and the colours are based on the commission's recommendations.

The pure spectral colours can be found on the curved line surrounding the tongue-shaped region of composite colours. The numbers along the curve represent corresponding wavelengths (in nanometres).

Figure 2 shows the chromaticity diagram with a few emission lines and bands of Table 2 drawn on the curve of spectral colours. The colours of the diagram are only approximate.

Figure 2. Chromaticity diagram with some emission bands. Click on the picture to see the true-colour version (44K jpg).

All would be well if we could just pick up the light from the above emitters. However, the emitting molecules, especially SrCl and BaCl, are so reactive that they cannot be packed directly into a firework. To generate them, we need pyrotechnic compositions designed to generate the above molecules, to evaporate them into the flame and to keep them at as high temperature as possible to achieve maximum light output. To get good colours, there must be substantial amounts of emitters present in the flame. The emitters are not alone: in order to achieve the high temperature, a fuel - oxidiser system is also needed, as well as some additional ingredients.

The colours of aerial fireworks come invariably from stars, small pellets of firework composition which contain all the necessary ingredients for generating coloured light or other special effects. They may be as tiny as peas or as large as strawberries. A typical red star might contain

Potassium perchlorate, 	67% by weight
Strontium carbonate	13.5%
Pine root pitch (fuel)	13.5%
Rice starch (binder)	6%

Care must be exercised in selecting the ingredients. The composition must be safe and stable in storage. In addition, it must work as expected and burn with a red colour once lit. For a deep red we need only SrCl and SrOH emission - and nothing else. To generate the emitting molecules at a sufficiently high temperature, a fuel-oxidiser system (pine root pitch - potassium perchlorate) is used. Strontium carbonate is used as the Sr source, and chlorine comes from potassium perchlorate (KClO4 --> K+ +Cl- + 2 O2). An excess of fuel is used to prevent the formation of SrO, which would solidify in the flame and emit grey body radiation. This will result in a "washed-out" colour. Too much fuel would be a disadvantage, too, because the glowing carbon particles quickly overwhelm the red colour.

Pure colours also require pure ingredients. Sodium D-line atomic emission is so strong and so easily excited that even minute amounts of sodium impurities will quickly ruin the colours. Potassium, with its weak atomic lines, does not interfere with most colours, and potassium salts can usually be used.

Organic fuels, such as pine root pitch, various gums and rosins and synthetic resins, cannot generate as high temperatures as metallic fuels. The pyrotechnist is tempted to use powdered magnesium and aluminium for his/her brilliant stars, because they provide an easy method of raising the flame temperature and increasing the brightness. Unfortunately, the molecular emitters are quickly destroyed if the flame is too hot. CuCl is probably the most fragile colour emitter. It can be used with metallic fuels only with difficulty. Consequently, blue stars are never very bright. Another problem with metals are their oxidation products, metal oxides, which are powerful grey body radiators due to their refractory nature. Their incandescent glow can easily wash out all colours.

Over the years, chemists, amateur pyrotechnists and professional fireworkers have solved most of the problems of coloured flame production. Excellent formulations exist for yellow, orange, red, blue and green stars. The problem I've been working on is the production of deep forest green or ocean green. As you can see in Figure 3., there are no bands in that region (490 nm - 500 nm). A composite colour made of BaCl and CuCl emissions is an obvious choice, but unfortunately BaCl emission is seldom - if ever - free from interfering BaOH and BaO emissions, which fall in the yellow and yellowish-green region of the visible spectrum. It seems that it is easier to generate greenish blue and turquoise than the long sought after bluish green and forest green.

Further reading:

  • John A. Conkling: Pyrotechnics. Scientific American, July 1990, 96.
  • Takeo Shimizu: Fireworks from a Physical Standpoint, Part II. Pyrotechnica Publications, 1983.
  • John A. Conkling: The Chemistry of Pyrotechnics. Marcel Dekker, Inc, 1985.
  • K.L. Kosanke: The Physics, Chemistry and Perception of Colored Flames. Part I: Pyrotechnica VII, 5 (1981). Part II: Pyrotechnica IX, 42 (1984). Pyrotechnica is a serial published by Pyrotechnica Publications.
  • Takeo Shimizu: Fireworks - The Art, Science and Technique. Second Edition. Pyrotechnica Publications, 1988.
  • For more information about the chromaticity diagrams and colour science, see: What is CD by Eugene Vishnevsky.

The heading photo of this article is from the 1994 PGI convention, in New Castle, PA, USA. Photo by Johan Heide.

Return to the pyrotechnics main page