Real-Time Measurements of Ozone and UV Radiation during Pyrotechnic Displays

Real-time measurements of ozone, NOx, aerosols and radiation were performed very close to the launch area of high intensity pyrotechnic spectacles, the so-called Mascletàs, typical of celebrations in eastern Spain. During these events, a considerable number of spectators are potentially exposed to high concentrations of pollutants directly or indirectly produced by the ignition of thousands of firecrackers at ground-level. This work is focused on the mechanism of ozone formation throughout the Mascletàs. After the initial decrease in O3 levels (minimum concentration < 10 μg m) due to the reaction with NO emitted by fireworks explosions, peak concentrations of up to ~150 μg m were recorded, clearly indicating that ozone was generated during these events. The results suggest that UV light produced by fireworks displays photolyzes O2 releasing O to form O3 as in the stratosphere.


INTRODUCTION
Several festivals worldwide are celebrated with pyrotechnic displays that are responsible for a significant increase in air pollutant concentrations.The burning of firecrackers and sparkles produces important amounts of particles (Moreno et al., 2007;Vecchi et al., 2008;Crespo et al., 2012;Lin et al., 2014) and gaseous pollutants, such as volatile organic compounds (Chang et al., 2011;Nishanth et al., 2012), sulfur dioxide (Moreno et al., 2007;Croteau et al., 2010;Chatterjee et al., 2013), nitrogen oxides (Moreno et al., 2007;Godri et al., 2010) and ozone (Attri et al., 2001;Nishanth et al., 2012;Kavouras et al., 2013;Yerramsetti et al., 2013).Although the deterioration of air quality during this type of events has limited time duration (generally less than 24 h), the potential health impacts cannot be considered negligible (Godri et al., 2010;Beig et al., 2013), especially for ground-level displays because of the proximity of the people attending the performance (Croteau et al., 2010).
Surface ozone is a secondary pollutant that can cause serious damage to human health, natural vegetation, crops and materials (Percy et al., 2003;Screpanti et al., 2009;Nuvolone et al., 2013;Tomer et al., 2015).Additionally, ozone acts as a potent greenhouse gas due to its light-absorbing properties (IPCC, 2007).Ground-level ozone is formed by chemical reactions of nitrogen oxides (NO x ) and volatile organic compounds (VOCs) in the presence of sunlight.The photolysis of NO 2 (λ < 420 nm) leads to the formation of atomic oxygen, which reacts rapidly with molecular oxygen to produce O 3 (Atkinson, 2000).
In the absence of other pollutants O 3 reacts with NO, resulting in not net formation or loss of ozone.
However, the chemical degradation of VOCs in the troposphere produces alkyl peroxy radicals (RO 2 ) and hydroperoxy radicals (HO 2 ) that can also convert NO to NO 2 , which obviously results in net formation of O 3 .

NO + RO
NO + HO 2 → NO 2 + HO (R5) A few previous studies have reported the generation of O 3 in the absence of sun radiation attributed to fireworks displays.Typical components of fireworks include an oxidizer (generally potassium nitrate), a reducing agent or fuel (commonly sulfur and charcoal), a binder used to improve the cohesion of the ingredients, and coloring agents (metals or metal salts).When metals are heated at the very high temperatures reached during fireworks ignition, they emit radiation covering a wide spectral range, including visible and UV light (Steinhauser and Klapötke, 2008;Russell, 2009).Light emission during fireworks displays can account for night-time formation of ozone, even when NO 2 is not present (Attri et al., 2001).However, mechanisms still remain uncertain.Attri et al. (2001) suggested that ozone is formed from the photolysis of molecular oxygen following a process equivalent to that occurring in the stratosphere.
Photodisociation of O 2 in the 176-242 nm range is the major source of stratospheric ozone (Parker, 2000): According to Attri et al. (2001) a fraction of the light emitted by the bursts of sparkles has a wavelength below 240 nm and thus is sufficiently energetic to dissociate the oxygen molecule and produce the atomic oxygen required for ozone formation.Alternatively, Nishanth et al. (2012) ascribed the formation of O 3 to reactions R1 to R5 since (1) there was no UV light in the emission spectra of different sparkles and powders they examined and (2) a positive correlation between O 3 and NO 2 was found.
In the present work, O 3 , NO, NO 2 , aerosols and radiation were continuously monitored during high intensity pyrotechnic events, the so called Mascletàs, which take place every year during the Hogueras de San Juan Festival in the city of Alicante (southeastern Spain).The main objective of this work is to improve the understanding of the mechanism of ground-level ozone formation during such episodes.

Monitoring Site
The Mascletàs are sonorous and visual pyrotechnic performances produced by the burst of thousands of firecrackers, principally at ground-level.These events take place every day at 2 pm from 19 to 24 June next to a downtown square of the city of Alicante (Fig. 1).A high number of spectators (around 20,000 people) gathers in the streets adjacent to the pyrotechnic venue and therefore are directly exposed to the pollutants produced throughout these events.
Measurements were made from a balcony (~15 m above ground level) of a building situated within the area where pyrotechnic devices were set off during the 2013 Alicante Festival (Fig. 1).It is important to indicate that the burst of firecrackers does not occur simultaneously in the entire area marked in Fig. 1.Moreover, the launch sequence of firecrackers varies from day to day, although the duration of each Mascletà is very similar (~9 minutes).For example, the event on 24 June can be divided in three stages of approximately three minute duration each one.The pyrotechnic spectacle began in the nearby square (zone A), in which aerial fireworks predominated.Then, firecrackers started being detonated, principally at ground-level, following the trajectory along the line from B to C. In the last stage, pyrotechnic explosions concentrated in zone C. Therefore, fireworks explosions are mobile sources relative to a fixed measurement location.A video of the Mascletà on 24 June is included in the Supplementary material.More information about the Mascletàs and the sampling location is given in Crespo et al. (2012).

Instrumentation
Ozone was monitored by UV absorption at λ = 254 nm using a 2B Technologies Inc. ozone analyzer, model 202.The instrument has 10-second measurement cycles with a dynamic range extending from a detection limit of 3.0 µg m -3 up to an upper limit of 1,000 µg m -3 .Data were registered every 10 seconds with a precision of 2%.
The measurements of NO x were carried out using a SIR chemiluminiscence NO-NO 2 -NO x analyzer, model S-5012.This technique relies on the measurement of light (λ = 600-2400 nm) emitted by excited NO 2 produced by the gas-phase reaction of nitric oxide and ozone.The intensity of light generated in the reaction is proportional to the concentration of NO present in the sample.NO 2 is indirectly measured by reducing it to NO onto a heated catalyst.The analyzer normal measurement cycle is approximately of 24 seconds with a 0.8µg m -3 detection limit and a precision of 0.5%.
Particle number size distributions in 31 size channels from 0.25 µm to 32 µm were measured using a Grimm 1109 optical particle counter, and data were recorded every 10 seconds.The uncertainty in the measurements of particle number concentration with a Grimm optical counter is lower than 10% taking a Mobility Particle Size Spectrometer containing a Differential Mobility Analyzer (MPS-DMA) as a reference for the same particle range (Burkart et al., 2010).
Radiation was measured every second with an AvaSpec-3648 fiber optic spectrometer with a scan range of 176-1100 nm and a bandwidth of 0.34 nm.The uncertainty in UV irradiance (200-240 nm) measurements is 10%.
The instruments were mounted on a platform and the monitor inlets were positioned facing the fireworks plume.Monitors started recording approximately 10 minutes before the beginning of the event, and recording continued for a further 5 minutes after it was finished.

RESULTS AND DISCUSSION
The amount of ozone detected at the monitoring site was dependent on the Mascletà design.Within the study period, ozone formation was recorded at the measurement point during the Mascletàs on 20 and 24 June.This work is mainly focused on the event on 24 June.

O 3 and Particles
Particles in the accumulation mode (0.1-1 µm) can be used as tracers of pyrotechnic activity (Wehner et al., 2000;  Nicolás et al., 2009;Richard et al., 2011;Crespo et al., 2014).In the present work, we have selected the particle number concentration (PNC) in the 0.25-0.28µm size channel as an indirect indicator of the arrival of the plume to the sampling site.Fig. 2 shows the variation in the concentrations of O 3 and particles between 0.25 and 0.28 µm during the Mascletà on 24 June 2013.
A sudden increase in aerosol concentration occurred just after the Mascletà started, accompanied by a rapid drop in ozone concentration from approximately 74 µg m -3 to less than 10 µg m -3 .During the whole event, which lasted approximately 9 minutes, an opposite variation in ozone and particle concentrations was found (r = −0.69,p-value < 0.05).This is likely due to the reaction between O 3 and NO, released along with particles from the combustion of fireworks, during the transport of the pyrotechnic plume to the measurement site.Additionally, calcium-rich particles resuspended during the displays (Crespo et al., 2012) could catalyze the heterogeneous destruction of ozone in a similar way to that reported during Saharan dust intrusions (Bonasoni et al., 2004;Nicolás et al., 2014).So, although it is obvious from Fig. 2 that ozone was generated during the Mascletà, the net effect of this event was a reduction in O 3 concentrations, as can be deduced from the data presented in Table 1.Average, maximum and minimum values correspond to concentrations measured for the period of the Mascletà.Background levels were calculated as the average of concentrations measured during 10 minutes before the start of the Mascletà.It is important to highlight that these are not the usual concentrations registered at the sampling site since the streets nearby the pyrotechnic site are closed to vehicular traffic 6 hours before the beginning of the displays.Data presented in Table 1 also show a spectacular increase in the concentrations of submicron aerosols and NO.Nitrogen oxides produced during fireworks activity arise from transformation of nitrate components of black powder (Drewnick et al., 2006).The increase in average NO concentration during the Mascletà is consistent with the observed reduction in mean O 3 concentration since a parallel increase in the levels of both pollutants is unfeasible.

O 3 and NO x
Fig. 3 shows the variation in the concentrations of O 3 , NO and NO 2 before, during and after the Mascletà.Abrupt changes in the concentrations of the measured species were observed throughout the event with a sharp decrease after the event was finished due to the quick dispersion of the pyrotechnic plume.
As mentioned before, at the beginning of the Mascletà O 3 was rapidly consumed by reaction with NO emitted by fireworks, resulting in the formation of NO 2 (reaction R3).However, the variation in ozone concentrations registered afterwards clearly indicates that ozone was generated throughout the event.
The absolute maximum (152 µg m -3 ) occurred approximately 4 minutes after the Mascletà started, and a secondary maximum (95 µg m -3 ) appeared around 2.5 minutes later.Both maximums were followed by a significant reduction in O 3 concentrations concurrently with a remarkable increase in NO levels (up to ~1500 µg m -3 ).Peak NO 2 concentrations occurred a little after maximums in NO.The most likely explanation for such a variation is that ozone was not formed from the reactions of NO x and VOCs, and that once O 3 was generated, it reacted with NO emitted by the successive bursts of firecrackers to produce NO 2 .In fact, the observed profile does not match with the typical diurnal pattern for ozone and NO x in polluted urban areas (Xu et al., 2011;Melkonyan and Kuttler, 2012;Hassan et al., 2013).
Fig. 4 shows the diurnal variation in the concentrations of ozone and nitrogen oxides in Alicante during the days leading up to the Festival.
The morning peaks in concentrations of NO and NO 2 are related to NO emissions during the rush hour traffic and the immediate conversion of NO to NO 2 as a result of the reaction with VOCs.As solar radiation increases ozone begins to accumulate, reaching a maximum during the central hours of the day due to higher photochemical activity.NO emissions during the evening rush hour cause ozone removal and therefore concentration decreases.It can be observed that the maximum concentrations of O 3 and NO x do not occur in the same order as in the Mascletà (see Fig. 3).

O 3 and UV Radiation
The obvious question that arises from the considerations above is: how is O 3 produced during these high intensity pyrotechnic displays?The most likely explanation is that atmospheric O 2 dissociates into atomic oxygen by absorption of UV light (λ < 242 nm) emitted during fireworks ignition.The subsequent combination of atomic and molecular oxygen enables ozone formation, as proposed by Attri et al. (2001).
In order to check that ozone was formed at ground level by reactions R6 and R2, it is required to confirm the emission of UV radiation during fireworks displays.The measurements made with the spectrophotometer corroborate that radiation containing the range of wavelengths capable of producing O 2 photodisociation was released throughout the Mascletà, as shown in Fig. 5. UV radiation intensity was integrated between 199.99 and 242.96 nm, then 4-second moving averages were calculated and normalized by the maximum value obtained.
The emission of UV radiation by the burst of pyrotechnic devices was observed during the whole event, although ozone formation was not always detected.The most important variables that influenced the detection of O 3 at the monitoring site were (1) the distance between the zone where fireworks explosions took place and the measurement point; and (2) the number of bursts per unit area, which could be called surface density of explosions.
At the beginning of the Mascletà (zone A, see section 2.1) a relatively low number of pyrotechnic explosions occurred far from the monitoring site.The observed reduction in the concentration of ozone could have been the result of the reaction between O 3 and NO emitted by fireworks during the transport of the plume to the monitoring site.
The principal maximum (zone B-C) was detected when a high number of fireworks explosions occurred very close to the sampling site (on the way from B to C).This outcome is consistent with the experiments by Attri et al. (2000) that showed a very good correlation between the quantity   In zone C a very high number of pyrotechnic explosions concentrated in a small surface area; however, the maximum O 3 concentration was lower than that registered at zone B-C since the distance from the measurement point was higher and a fraction of the O 3 generated by the photolysis of O 2 could have been consumed by reactions during the transport of the pyrotechnic plume.The correlation coefficient between ozone concentration and UV radiation intensity for the interval corresponding to this maximum was 0.85 (p-value < 0.05) excluding the time delay (Fig. 6).Temporal variations in the concentrations of the measured pollutants over the Mascletà on 20 June were similar.Two maxima of ozone concentration and UV radiation intensity were also recorded.The correlation coefficients between both variables were 0.75 (p-value < 0.05) for the principal maximum and 0.80 (p-value < 0.05) for the secondary maximum.

CONCLUSIONS
During the summer festival held annually in Alicante (southeastern Spain), thousands of firecrackers are set off, primarily at ground level, in a short and intense pyrotechnic spectacle (Mascletà) launched every day at 2 pm in the city center.Ozone, nitrogen oxides, aerosols and radiation were continuously monitored throughout these events in 2013 with the purpose of elucidating the mechanism of O 3 formation associated to fireworks displays.Despite the fact that ozone average levels for the interval of the Mascletà (~9 min) were lower than those measured before the beginning of the event due to reaction with NO emitted by pyrotechnic explosions, peak O 3 concentrations of about 150 µg m -3 were observed.Our data clearly suggest that ozone was unlikely formed from reactions involving nitrogen oxides.Instead, the photolysis of oxygen molecules by absorption of UV radiation emitted by firework displays, and the subsequent combination of atomic and molecular oxygen, is the most plausible mechanism to explain ozone formation.This reaction sequence corresponds to the process of ozone production in the stratosphere.The results that support the proposed mechanism are: (1) the temporal variation in the concentrations of O 3 and NO x during pyrotechnic events was different from the usual daily cycles observed in the troposphere; (2) UV light in the range of wavelengths capable of producing the photolysis of O 2 was detected throughout the whole pyrotechnic spectacle; and (3) a significant positive correlation between ozone and UV irradiance was obtained for the intervals in which ozone concentrations were maxima.

Fig. 1 .
Fig. 1.Location of the measurement site (indicated by a star) and the Mascletà area.A, B and C are related to the sequence of the Mascletà.

Fig. 2 .
Fig. 2. Temporal variation of ozone and particle number concentrations (PNC) in the 0.25-0.28µm size range during the Mascletà on 24 June 2013.Lines indicate the start and end of the pyrotechnic event.

Fig. 4 .
Fig. 4. Typical daily cycles of ozone and nitrogen oxides in Alicante during summer.

Fig. 5 .
Fig. 5. Temporal variation of ozone, particle number concentrations (0.25-0.28 µm) and UV radiation during the pyrotechnic displays on 24 June 2013.The different areas marked in the figure correspond to the zones where fireworks are successively ignited (see section 2.1 and video in the Supplementary material).

Fig. 6 .
Fig. 6. (Left) Temporal variation of ozone and UV radiation during the pyrotechnic spectacle on 24 June 2013 after applying the time delay to UV radiation.(Right) Correlations of ozone with UV radiation for the intervals of the two maxima.

Table 1 .
Concentrations of gaseous pollutants and particles (PNC) in the 0.25-0.28µm size range before (background) and during the Mascletà on 24 June (average, maximum and minimum).
Fig. 3. Temporal evolution of ozone and nitrogen oxides concentrations during the pyrotechnic performance on 24 June 2013.Lines indicate the start and end of the pyrotechnic event.