Influence of atmospheric conditions

Methods

The typical measurement cycle was 4 min, divided into two sampling periods of 2 min each: close to the filaments (inside the chamber) and outside the chamber, the latter being representative of background atmosphere. Mean particle concen-trations in these two conditions, respectivelyN_in andN_out, were obtained by aver-aging the raw number concentrations over the second minute of the corresponding period, in order to allow for device stabilization. Following the same strategy as for differential pump-probe LIDAR measurements (see Sec. 3.1.2), we considered the triple

N_out,i−1,N_in,i,N_out,i , where the indexiidentifies one single measure-ment cycle. The mean relative increase of particle concentration was defined as follows:

E_i=N_in,i−1

2(N_out,i−1+N_out,i). (3.2)

This corresponded to one individual data point in the subsequent data process-ing, representing the laser-assisted particle generation as the difference between particle density close to the filament and the average of the preceding and the

fol-Figure 3.10: Correlation of temperature, relative humidity and water vapor vol-ume mixing ratio with the laser-induced particle increase for each size class. [19]

lowing background reference, in order to better account for potential atmospheric long-term drifts.

Observations

Atmospheric conditions had specific and contrasting influences on the laser-assisted yield of different particle sizes. The correlation between atmospheric parameters and the particles generation is illustrated in Fig. 3.10. RH was positively cor-related with the generation of particles below 400 nm, and negatively corcor-related above, while the opposite correlations were observed for temperature. Conversely, water vapor volume mixing ratio (VMR) was anti-correlated with respect to par-ticles under 500 nm, positively correlated with parpar-ticles above 3 µm, and uncor-related with the generation of particles in the 600 nm−3µm range.

This contrasted behaviour defined three particles size ranges:

• Nanoparticles (25 nm median diameter, as compared with 55 nm in the background), whose concentration was increased by the laser under all the conditions encountered during the campaign. Their increase was positively correlated with both relative and absolute humidity (Fig. 3.11b,c), and neg-atively correlated with temperature (Fig. 3.11a).

• Particles in the 230−400 nm range featured the following regime: the effect

3.2 Direct characterization of water vapor condensation in real atmosphere

Figure 3.11: Effect of laser filaments on the number density of nanoparticles (25−300 nm, panels a-c), 325 nm (panels d-f), 540 nm (panels g-i), 1.15 µm (panels j-l), as a function of temperature (a,d,g,j), RH (b,e,h,k) and water vapour concentration (c,f,i,l). Individual data points are calculated according to the definition 3.2. Error bars are obtained by sorting all individual measurements by ascending order of the considered atmospheric parameter, and computing one standard deviation for successive blocks encompassing 5% of the measurements.

Dotted lines display linear fits of the data. [19]

of the laser decreased with increasing absolute humidity (Fig. 3.11f) and temperature (Fig. 3.11d). Conversely, it was negligible below 50% RH, and increased when RH rises up to 100% (Fig. 3.11e).

• The dependences featured by the second particles size range vanished for diameters close to 500 nm (Fig. 3.11g,h,i), and reversed above, thus defin-ing the third regime (Fig. 3.11j,k,l). Note, however, that the increase of micrometer-sized particles rised again when approaching 100% RH.

Trace gases

Large amounts of O₃ and NO₂ were generated by the filaments. Typical con-centrations were, respectively, 200 ppb and 25 ppb during laser operation, and were independent of temperature and humidity (Fig. 3.12a,b). Note that the typi-cal atmospheric content of these species is one order of magnitude lower. As for the aerosols, the same consideration on the dilution factor between the filament volume and the chamber holds for gases. Thus, concentrations in the laser ac-tive volume were∼10⁵ times higher. Fig. 3.9b displays the decay time of these

Figure 3.12: Laser-induced concentrations of ozone and NO₂as a function of (a) temperature and (b) relative humidity. [19]

3.2 Direct characterization of water vapor condensation in real atmosphere

gases, after the laser was switched off. Time constants for gases and particles are comparable, suggesting that the decay was mostly driven by dilution through the chamber openings and owing to the airflow induced by sampling devices. While the ozone concentration decreased to its background level in 10 min, NO₂featured a retarded effect typical of a secondary product generated by the oxidation of NO by ozone [6], thus also speeding up the decay of the latter. The same behaviour was reproduced over more than 30 experimental realizations.

Interpretation

The observations on the generation of aerosols as a function of atmospheric pa-rameters clearly evidenced three different regimes, for nanometric, sub-micronic, and micrometric scale particles. These experimental results suggest that the fol-lowing three-steps condensation mechanism could describe the observed laser-assisted particle concentration increase and particle size growth.

In this scenario, the laser acts as a ’trigger’, initiating particle formation via dif-ferent reaction mechanisms, as will be illustrated in Sec.4.2. As a result, parti-cles with a median diameter of ∼25 nm accumulate inside the filaments. The subsequent behaviour of the laser-generated aerosols is completely driven by at-mospheric conditions. Indeed, these particles can grow if the RH is sufficient to ensure their stability; if this is not the case, they will evaporate immediately. Once they reach∼500 nm, which is typically the size of medium cloud condensation nuclei [6], their evolution is expected to be limited by the concentration of water molecules, since water condenses from the gas phase at their surface. This may explain why the growth of these particles requires a high VMR, and hence higher temperature, as illustrated by the correlation of these parameters with big particles in Fig. 3.10. Conversely, the depletion of the humidity by particle water uptake results in less water in the gas phase, and hence the negative correlation between RH and micrometric particles. It is important to note that water uptake also oc-curs in the preceding step, when nanometric particles start to grow; one should then somehow expect a positive correlation with VMR also in this case, despite the fact that Fig. 3.11f shows an opposite trend. However, it is well known from Kohler theory (see Sec. 2.5.2) that small particles stability is highly dependent on relative humidity, while this feature vanishes with increasing diameters. We can conclude then that relative humidity rather than absolute water content is the main limiting factor for particles to grow in this initial phase. As a droplet increases in size, it requires bigger and bigger amounts of available water, hence the depen-dence on RH and VMR reverses.

This scenario requires a highly efficient stabilizing mechanism to prevent the par-ticles from re-evaporating well below 100%RH. The standard Kohler theory al-lows for a qualitative interpretation of the observed trends in particles growth as

a function of atmospheric parameters. Nevertheless, it states that the stability of particles smaller than 1 µm requires a large supersaturation, which would be totally unrealistic in standard sea-level atmosphere. Our observations, however, clearly show that stable nanometric particles are formed under the action of the laser, even in sub-saturated atmosphere.

The mechanism that makes this possible may be provided by the strong impact of the laser filaments on the local chemical composition of the atmosphere. Mea-surements of nanoparticles and trace gas concentrations, assuming a dilution fac-tor of∼10⁵between filament volume and protection chamber, lead to an estimate that up to 10¹¹ ozone molecules (8×10⁻¹¹ g) and 2.5×10¹⁰ NO₂ molecules (2×10⁻¹¹ g) were available for each nanoparticle. These amounts could allow an efficient chemistry of nitrogen oxides, eventually leading to the formation of very high concentrations of highly hygroscopic nitric acid, typically 1000 times the ppb level at which HNO₃ is known to stabilize water droplets in an atmosphere slightly below 100% RH [6]. We can therefore expect efficient stabilization at much lower RH through binary HNO₃−H₂O condensation.

The need for a certain nitric acid content inside each droplet also represents a limiting factor for particle growth towards the tens of micrometers range, where indeed no particles were found. While water vapour is highly abundant in the air, HNO₃ was only injected in the atmosphere by the laser, and thus in limited amounts, even if high with respect to standard concentrations (see Sec. 2.3.5).

In order to assess the validity of this tentative explanation of the condensation pathway, we performed complementary laboratory tests, oriented rather on the chemical composition of generated aerosols, which will be discussed in Sec. 3.3.

Additionally, we proposed a theoretical model based on a modified standard Kohler theory, to account for the extremely high nitric acid concentrations within the fil-ament volume, described in Sec. 3.4.