Figure 1 shows a schematic representation of the SPAMS ion source. The Nd:YAG laser beam is focused in the center of the ion source where particles are ionized by laser radiation. The 266 nm laser operates in TEM00 mode with a Gaussian distribution. The e-1 width of the focused Gaussian laser beam is about 300 µm. Factors that actually affect the ionization efficiency are the total laser pulse energy and the location of the particle in the focused laser spot. As the particle size is more than 300 times smaller than that of the focused laser beam width, each particle is ionized by a virtually uniform laser power density. But when the beam of the particles is wider than the diameter of the laser beam in the ionization region, the particles in the beam can be exposed to very different laser energy fluencies, resulting in a great difference in ionization efficiency. Therefore, when the particles travel near the edge of the laser spot, it becomes impossible to generate enough ions to detect such particles.

In our experiments, the Ultra compact Q-switched Nd:YAG laser (Quantel Laser) was used with a pulse width of 7.2 ns. The pulse energy of the laser was set to 0.6 mJ, which corresponds to a laser power density in the focal point of about 2×108 W cm-2.

When the delayed extraction technique is not used, the two DC high-voltage potentials are constantly applied to the extraction electrodes of the original bipolar SPAMS ion source. When the charged particles enter the space of the ion source, they start to be deflected under the action of the DC electrical field. The amount of deflection is related to the speed and the mass of the particles, the amount of the surface charge, and the electrical field strength. Thus, during the flight through the ion source, the particles gradually deviate from the laser center point, resulting in the decrease in the ionization efficiency and a significant drop in the hit rate.

When the delayed extraction technique is used, the particles enter the extraction region while the electrical field is kept at zero. Then the laser ionizes the particle, and the high voltage extraction pulse is applied with 100 ns delay after the laser pulse. So there is no electric field between the ion source plates before the particles are ionized by the laser. Thus, the charged aerosol particles are not deflected by the electric field force during their flight, thereby achieving a higher hit rate.

The influence of the electric field on the hit rate was studied by using PSL beads of five different particle sizes. In Fig. 2 the dependencies of the hit rate on the electrical field strength for DC and DE modes are shown. It can be seen that under the same DC electric field strength (Fig. 2a), the hit rate of the particles from 320 to 1400 nm increases, indicating that the effect of the electric field on the small particles is more pronounced. As the DC electric field strength increases, the hit rate of the particles of each size gradually decreases, indicating that the particles are deflected by the electric field force.

Under the DE mode, the hit rate for the particles of all sizes is obviously increased in comparison with the DC case (Fig. 2b). When the pulsed electric field strength exceeds 60 kV m-1, the hit rate of the particles above 520 nm is close to 100 %, indicating that the flight path without an electrical field is essential for the increase in the hit rate of the particles. The hit rate of the 320 nm PSL particles is 40 %, which is much lower than 100 % of other particles. This can be due to the focusing effect of the aerodynamic lens. The divergence of the particle beam itself is higher than that for the particles above 500 nm, which results in a decrease in the hit rate. In addition, when the electric field strength increases from 0 to 60 kV m-1, the hit rate gradually increases from 5 % to 100 %. We speculate that the increase in the hit rate with the electrical field strength in this range may be caused by the increase in the resolving power of the TOF MS (Cotter, 1994). The lower the extraction electrical field strength is, the bigger the turnaround time in the TOF MS is, which in turn decreases the resolving power of the instrument. In the case of a large turnaround time, the peak width increases and the peak height decreases. This, in turn, can result in a loss of measured ion current because of an ADC threshold adjusted to reject the low-intensity peaks accounted as noise. The loss of the recorded ion current can in turn result in a miscount of some particles, thus decreasing the hit rate. Also, for the low extraction electric field, ion divergence can be enhanced, resulting in a loss of ion current during the flight of the ion to the detector.

In Fig. 3 dependencies of the hit rate gain achieved for the DE mode compared to the DC mode case are shown. It can be seen that for different particle sizes, the hit rate gain increases sharply with the increase in the electric field strength, especially for small particles. For 320 nm particles, the hit rate can be increased by more than 100 times at an electrical field strength over 180 kV m-1. This result shows that the DE technique can effectively eliminate the influence of the electric field intensity on the flight path of the particles, avoid the use of additional peripheral devices such as electrostatic neutralizers, and result in an improvement in the hit rate.

It is worth noting that the mass spectra obtained with DE have a higher resolution, but they are approximately the same set of peaks as those obtained with the injection of ions in a constant field (CF). Peaks in the mass spectrum obtained with DE are narrower and, on average, have a larger amplitude. The influence of a ∼100 ns delay is not reflected in the composition of peaks in acquired mass spectra. Mass spectra of PSL particles acquired in DE and DC modes are presented in the Supplement for comparison. The mechanisms of ion formation as part of a complex of processes during laser exposure to a particle are of great interest and are the subject of our further research. In this work we did not focus on these issues, since they are not directly related to the observed increase in the hit rate.

In order to further confirm that the particles are deflected by the action of the electric field, we conducted several further experiments. Taking the 720 nm PSL microsphere as an example, we compared the hit rate of 720 nm PSL beads under the constant DC electrical field only to the hit rate of those under the DC electric field with an electrostatic neutralizer and to the hit rate under the delayed extraction combined with or without a neutralizer. Figure 4 shows the hit rate value for a measurement period of ∼30 min. It can be seen that the particle hit rate after the use of the electrostatic neutralizer has increased from an average level of 20 % to a level exceeding 85 %, indicating that the surface charge of the particles has a great influence on the hit rate. However, the average hit rate of the particles obtained by the DE technique is close to 96 %, which is slightly higher than that of the hit rate measured in the case when the electrostatic neutralizer was used. This difference can be caused by a residual charge remaining on the particles even after the usage of an X-ray neutralizer (TSI 3088). The higher hit rate achieved by delayed extraction is caused by the absence of electric field influence on the particle motion.

A further two experiments were performed to measure particle beam divergence under the influence of the electrical field formed between two electrodes of the extraction region of the SPMS. The particles were collected on the surface of the glass plate, as it is shown in Fig. 1. In the first experiment (Fig. a, b) no voltage was applied to the electrodes, and the resulting spot just represents a 720 nm particle beam profile as a result of the aerodynamic dynamic lens action. The profile was fitted by a Gauss envelope, and the average beam width extracted from the Gaussian curve is a little more than 0.1 mm. The second experiment (Fig. 6a, b) was performed with DC potentials (+/-1000 V) applied to the extraction region electrodes, in accordance with the usual working conditions of the DC extraction mode. The trace of the same 720 nm PSL beads on the glass plate is shown in Fig. 6a.

The particle divergence is obviously increased in this case. Using the ImageJ program, the trace profile was extracted, and then the Gaussian fitting was uploaded on the data. The space distribution of the particles can be easily transformed into a particle charge distribution as we deal with the calibrated PSL beads. The ion beam displacement at the exit of the extraction region (Fig. 6a, b) was analyzed by Liu et al. (1995), and it can be described by the simple equation: 1Δx1=12at2=12qmVHmpdL1Up2. Here q and mp are particle charge and mass, d and VH are the distance and voltage difference between two flat extraction electrodes, L1 is the electrode height, Up is the particle longitudinal velocity, a is the acceleration of the particle caused by the extraction electrical field, and t is the particle dwell time in the extraction region. Thus, the particle trajectories are parabolic, scaled by their mass and charge. The equation was derived from a simplified assumption that the particle moving between the plates is accelerated by the constant horizontal electrostatic force. Then the particle with the horizontal speed component moves until it hits the glass gathering plate. The resulting particle displacement in the position of the gathering glass plate is as follows: 2Δx=Δx1+Δx2=12qVHmpdL1L1+2L2Up2. Here L2 is the distance from the lower edge of the extraction electrode to the gathering glass plate surface. The resulting particle displacement is proportional to the particle charge; so from the trace in Fig. 6a we can extract the particle charge distribution shown in Fig. 6b. The average particle charge (in units of e) extracted from the Gaussian envelope is 60 elementary charges. There are multiple publications in which this electrification charge was measured for different types of particles (Dodd, 1953). The results reported in the literature (40 charges per 700 nm road dust particle; Forsyth et al., 1998; 40 charges per 400 nm PSL particle; Kleefsman et al., 2008) are in reasonable agreement with our estimation made for 720 nm PSL beads.

Using known particle mass, velocity, extraction voltage, and electrodes dimensions, we have extracted the particle displacement per unit charge in the middle point on which the ionizing laser is focused. For the parameters of our instrument this displacement is 12.5 µm z-1, where z is the number of elementary charges on the particle. Hence, a particle with an average of 20 elementary charges will be deflected ∼250 µm from the ionizing laser spot center. Taking into account the Gaussian profile of the focused laser beam of 300 µm, the hit efficiency will be greatly affected by such a displacement.

The comparison of the detection efficiency for the instrument with DC extraction and with DE used for the actual atmospheric particulate matter detection is shown in Fig. 7a and b. Aerosol particles from laboratory room air were analyzed by the SPMS used in this work. In order to ensure that aerosol composition is stable during the experiment, the DC and DE modes of operation were switched alternately after 10 min. Four sets of both DC and DE extraction data were acquired in 80 min. The hit rate was calculated as the ratio of the number of ionized particles to the number of sized particles. The hit rate data were extracted for the groups of particles selected by size; bin value was 20 nm while the whole range was 200–1000 nm. The data of four sets of 10 min each, both for DC and DE extraction modes, were summed. In the DC mode, in 40 min, a total of 12 174 particles was recorded, of which 4569 were ionized, while in the DE mode, 12 030 particles were recorded, and 8228 were ionized for the same time. It is seen from Fig. 7a that the center of the distribution obtained with DC extraction (blue circles) is shifted to the higher particles size. This behavior corresponds to our results obtained for the model PSL particles. Figure 7b shows the hit rate gain from the DE technique to the DC extraction, obtained from the data plotted in Fig. 7a. It should be noted that usage of delayed extraction eliminates the distortion of particle distribution, which results from the dispersion of particle deflection by their size in the case of DC extraction. The integral hit rate for the experiment with the ambient particles is ∼25 % higher in the case of delayed extraction (Fig. 7a), while the gain in hit rate achieved for small particles is ∼4 times bigger than that measured for DC extraction (Fig. 7b). The increased effect for small particles can be explained by Eq. (1). The smaller the m/q of the particle, the bigger its deflection in the ion source (extraction region). Hence, the hit rate of smaller particles is decreased more, and the gain caused by DE should be higher.

The hit rate of the atmospheric aerosol is lower than that of the standard PSL beads mainly due to the shape of the particles. In the actual atmosphere, the shape of the particles is much more complicated than that of the PSL sphere. For example, the aged carbon particles tend to be spherical, and the hit rate is relatively high, while the fresh black carbon particles exhibit a chain structure, so the divergence during their flight is probably bigger, resulting in lower hit rates (Ghan et al., 2012). In general, using the delayed extraction technique, the hit rate of environmental particles with a size of <500 nm can be noticeably improved, and the obtained particle size distribution becomes closer to the real one.