Single Charging of Nanoparticles by UV Photoionization at High Flow Rates

Abstract

The feasibility of UV photoionization for single unipolar charging of nanoparticles at flow rates up to 100 l· min ⁻¹ is demonstrated. The charging level of the aerosol particles can be varied by adjusting the intensity of the UV radiation. The suitability of a UV photocharger followed by a DMA to deliver monodisperse nanoparticles at high aerosol flow rates has been assessed experimentally in comparison to a radioactive bipolar charger ( ⁸⁵ Kr, 10 mCi). Monodisperse aerosols with particle sizes in the range of 5 to 25 nm and number concentrations between 10 ⁴ and 10 ⁵ cm ⁻³ have been obtained at flow rates up to 100 l· min ⁻¹ with the two aerosol chargers. In terms of output particle concentration, the UV photoionizer performs better than the radioactive ionizer with increasing aerosol flow rate. Aerosol charging in the UV photoionizer is described by means of a photoelectric charging model that relies on an empirical parameter and of a diffusion charging model based on the Fuchs theory. The UV photocharger behaved as a quasi-unipolar charger for polydisperse aerosols with particles sizes less than 30 nm and number concentrations ∼10 ⁷ cm ⁻³ . Much reduced diffusion charging was observed in the experiments, with respect to the calculations, likely due to ion losses onto the walls caused by unsteady electric fields in the irradiation region.

INTRODUCTION

Particle size selection is a crucial step in aerosol techniques for the synthesis of nanoparticles and of nanostructured particulate materials with tailored properties for use in a wide range of functional applications (Kruis et al. 1998; Wegner et al. 2006). Nowadays, electrical mobility analysis is a worldwide used technique to control particle size in gas phase synthesis of nanomaterials under atmospheric pressure conditions in the laboratory (Maisels et al. 2002a; Nasibulin et al. 2005; Awano et al. 2006; Kim and Zachariah 2007). Basically, the aerosol passes first through a device where the particles acquire a charge distribution (aerosol charger) and then enters a Differential Mobility Analyzer (DMA) where particles are classified based on the electrical mobility, which is a function of the particle size, charge, and shape. The size distribution of the particles selected with the DMA depends upon the particle charge distribution of the polydisperse aerosol upstream and upon the resolution of the DMA. Present aerosol systems for in-flight size selection of nanoparticles rely on diffusion charging, mostly radioactive bipolar chargers and corona unipolar chargers, and on nano-DMAs with optimized resolution and transmission for particles in the nanometric range (Chen et al. 1998). In that way, nanoparticles of a well-defined size with geometric standard deviation σ _g < 1.1 and number concentrations up to 10⁵ cm⁻³ are delivered at gas flow rates of a few liters per minute.

The applicability of electrical mobility analysis to commercial production of size-selected monodisperse nanoparticles in gas phase at atmospheric pressure demands the development of new aerosol chargers and DMAs capable to cope with the much higher flow rates used in aerosol processes for the fabrication of nanomaterials in the industry, in the order of m³· min⁻¹ and higher. Electrical mobility analysis for size selection of nanoparticles at gas flow rates up to 10² l· min⁻¹ has been demonstrated in a recent work (Hontañón and Kruis 2007). By using an aerosol charger based on a β -source of ⁸⁵Kr (10 mCi) followed by a DMA that operates at high aerosol flow rates (HF-DMA), monodisperse nanoparticles (d _p < 25 nm) with number concentrations of 10⁴ to 10⁵ cm⁻³ were delivered at a flow rate of 10² l· min⁻¹.

The main advantage of diffusion charging is that it depends only weakly upon aerosol material (Davison and Gentry 1985). By bipolar diffusion charging, particles less than 30 nm acquire at most one charge, which is a major requirement for application to the production of monodisperse nanoparticles; nonetheless, single charging efficiencies are rather low (Wiedensohler 1988). Due to the absence of ion recombination, unipolar ionization results in higher ion concentrations than bipolar ionization. As a consequence, unipolar charging is largely more efficient, in terms of total fraction of charged particles and particle charge level, than bipolar charging. For application to size selection of nanoparticles, however, unipolar aerosol chargers must lead to acceptable fractions of singly charged particles with minor fractions of multiply charged particles in the aerosol at the outlet and minimal particle losses within the charging device.

Radioactive materials are reliable and stable sources of bipolar ions. They are used in bipolar and unipolar aerosol chargers. In the latter case, ions of the desired polarity are attracted to and mixed with the aerosol, whereas ions of the opposite polarity are collected onto an electrode without encountering the aerosol (Adachi et al. 1992; Wiedensohler et al. 1994; Chen and Pui 1999). On the other hand, diffusion charging of nanoparticles by soft X-ray photoionization has been found suitable for aerosol measurement purposes (Shimada et al. 2002; Han et al. 2003; Lee et al. 2005). The use of radioactive and X-ray sources in the industry, however, is not advantageous because of large added costs and safety issues. Maintenance and replacement are also motive of concern mainly for X-ray sources, because of the limited life-time of the lamp and/or the window through which the radiation enters the aerosol charging region.

Corona discharge is another source of unipolar ions. Corona unipolar diffusion charging of nanoparticles has been widely studied (Büscher et al. 1994; Kruis and Fissan 2001; Biskos et al. 2005; Alguacil and Alonso 2006). Both nanoparticle charging efficiency and transmission have been assessed in dependency of particle size in experiments with monodisperse aerosols having low particle number concentrations (< 10⁵ cm⁻³) and typical flow rates in DMAs (1–2 l· min⁻¹). Data on charge distribution of nanoparticles in corona ionizers are, however, scarce. There is no data on corona diffusion charging of polydisperse aerosols in the nanometric range with high particle concentrations either. On the other hand, corona discharge can promote the formation of new particles, either from erosion and sputtering from the corona electrode itself or from gaseous contaminants present in the system; the presence of water being a decisive factor in the particle generation process (Alonso et al. 2006). Therefore, the suitability of corona discharge as a stable and clean source of unipolar ions for long-term continuous charging of aerosols needs further assessment.

Aerosol charging by UV irradiation has been investigated in the past (Jung et al. 1988; Matter et al. 1995; Mohr et al. 1996). Under exposure to UV radiation, electrons are emitted from the surface of the aerosol particles, which acquire positive charges, while photoelectrons attach instantaneously to gas molecules and form negative ions. Positively charged particles discharge by attachment of negative ions to their surface and eventually become negatively charged. Hence, by UV irradiation particles can be unipolarly positive or bipolarly charged depending upon the ion concentration in the gas, which increases with the particle number concentration. In general, the particle charge distribution shifts from unipolar to almost symmetrically bipolar with increasing aerosol particle concentration (Mohr and Burtscher 1997; Maisels et al. 2003). Aerosols can be unipolarly charged if negative ions are removed from the aerosol in times shorter than the time for the ions to diffuse to the particles. Contrarily to diffusion charging, photocharging depends strongly upon particle composition. In order to escape, the kinetic energy of the electron perpendicular to the particle surface must be larger than a given threshold, which depends upon the work function of the material and upon the particle size.

This work explores the feasibility of UV irradiation for single unipolar charging of nanoparticles at high flow rates. A photocharger emitting UV light with a wavelength of 172 nm and variable intensity is used for that purpose. The performance of the UV photocharger is studied experimentally and theoretically in dependency of particle size, the particle number concentration, the radiation intensity and the irradiation time. Experiments are carried out with aerosols of tin oxide having particles in the size range of 3 to 30 nm, number concentrations ∼10⁴ cm⁻³ (monodisperse aerosols) and ∼10⁷ cm⁻³ (polydisperse aerosols), and flow rates between 30 and 100 l· min⁻¹. Aerosol charging in the UV photoionizer is described by means of a model including photocharging and diffusion charging, and an analytical solution is derived for the case when diffusion charging is negligible. The model of photoelectric charging depends upon an empirical parameter which is determined by fitting of the analytical solution to the experimental data obtained with monodisperse aerosols, and the values so obtained are used to simulate charging of polydisperse aerosols. Aerosols with particle number concentrations as high as 4· 10⁷ cm⁻³ are charged almost unipolarly in the experiments by adjusting the intensity of the UV light according to the aerosol flow rate; while the simulations showed a much larger diffusion charging effect. Mechanisms other than diffusion could have contributed to ion depletion inside the photocharger, likely electric fields induced by the alternating high voltage applied to the UV lamps.

THEORETICAL MODEL

Modelling of aerosol charging by simultaneous photoionization and diffusion of gas ions has been addressed in previous works (Maisels et al. 2002b; Jiang et al. 2007a). The evolution with time of the concentration of ions and particles in an aerosol exposed to UV radiation is governed by the population balance equations

where N _i is the concentration of negative ions in the gas; ω _i is the ion depletion rate to the walls; N _p is the total aerosol particle concentration; and, for each particle size class s and particle charge class q: N _s,q is the particle concentration, α _s,q is photoionization rate, η _s,q is the ion-to-particle-attachment rate coefficient, and ω _s,q is the particle depletion rate to the walls.

The ion-to-particle attachment coefficients are calculated on the basis of the theory of the limiting sphere by Fuchs (1963), as follows

where δ denotes the radius of the limiting sphere, which is a function of the particle radius a and of the ion mean free path λ _i ; c _i and D _i are, respectively, the mean thermal velocity and the diffusivity of the ion; and γ is the collision probability (Hoppel and Frick 1986), which depends on the electrostatic potential energy φ of the ion in the field of the particle. Given the electrical mobility Z _i and the mass m _i of the ions, the diffusivity D _i , the mean thermal velocity c _i , and the mean free path λ _i can be determined by using well-known relationships from the kinetic theory of gases (Liu and Pui 1977; Adachi et al. 1985). The potential energy of the ion in presence of a sphere is the sum of the Coulomb φ _C and the self-image φ _i terms. The former is given by ϕ _C (r) = ± e ²/4π ϵ _o | q|/r and the later by ϕ _i (r,ϵ _r ) = e ²/4π ϵ₀ (− 1/2 ∑ _{k = 1} ^∞ k (ϵ _r − 1)/k (ϵ _r + 1) + 1 a ^{2k + 1}/r ^{2k + 2}) (Burko 2002; Messina 2002; Rinke et al. 2004); where a and q are, respectively, the radius and the charge of the sphere, r is the distance from the center of the sphere, e is the elementary charge (1.60· 10⁻¹⁹ C), ϵ _o is the permittivity of vacuum (8.85· 10⁻¹² C²· N⁻¹· m⁻²), and ϵ _r is the dielectric constant of the sphere.

On the other hand, photoelectrons are emitted from the surface of a sphere at a rate

where Y(hν) denotes the electron yield per incident photon, hν is the photon energy, and I is the power density or intensity of the radiation. Current models on photoelectric aerosol charging rely on the Fowler-Nordheim law for photoemission from clean surfaces Y(hν) = C (hν − Φ) ^m (Fowler 1931); where C and m are material-dependent constants and Φ stands for the photothreshold. For metallic spheres, Φ has the form Φ = Φ_∞ + e ²/4π ϵ _o (q + 1/a − 5/8a) (Wood 1981); where Φ_∞ denotes the work function, i.e., the photothreshold for an infinite planar surface, which is a characteristic of the material.

The current data base on photoemission of nanoparticles is rather scarce and focuses mostly on metals. The Fowler law has been successfully used to predict the photoemission yield of a variety of metallic particles (Ag, Cu, Pd, Au) in the vicinity of the threshold, hν − Φ ≤ 1.5 eV, with m = 2. The photoemission constant C, however, is a major uncertainty of the model. It has been observed experimentally that C is larger for particles than for macroscopic surfaces (Burtscher et al. 1982; Schleicher et al. 1993) and the enhancement factor depends upon the particle size and upon the light absorption behavior of the particle. It was found that the photoemission constant of Ag particles in air increased in a factor of 4 when decreasing the particle radius from 3 to 2 nm (Schmidt-Ott et al. 1980). On the other hand, C was approximately constant in experiments with Ag particles of radii between 2.7 and 5.4 nm in helium (Müller et al. 1988). In experiments in which particles of Ag and sucrose were exposed to soft X-ray radiation in presence of nitrogen, C increased in a factor of 2 when reducing the particle diameter from 15 to 6 nm for both aerosol species (Jiang et al. 2007b).

In experiments on aerosol charging by light irradiation both the photoemission constant C and the intensity of the incident light I are commonly unknown, so that the parameter K = CI is introduced, which is determined empirically (Maisels et al. 2002b, 2003; Jiang et al. 2007a, 2007b). After substituting in Equation (3), the photoionization rate α is expressed as α = K π a ²/hν [hν − Φ_∞ − e ²/4π ϵ _o a (q + 3/8)]². This equation shows that, when hν > Φ, a sphere can be photoionized up to a maximum charge level at which the process saturates (Matter et al. 1995); the saturation charge is given by q _max = 4π ϵ _o /e ²(hν − Φ_∞) a − 3/8.

Solution when Diffusion Charging is Negligible

The limiting case in which photoionization dominates over ion attachment to particles, α ≫ η N _i, is analyzed here. The ion balance equation is irrelevant in this case. We assume for simplicity that the aerosol is monodisperse and wall losses are neglected. In addition, we introduce the nondimensional variables f _q = N _q /N _p , τ _q = (α _q t _r )⁻¹, with 0 ≤ q ≤ q _max, and τ = t/t _r , where t _r denotes the mean residence time of the aerosol in the irradiation region. Then, the particle balance equations reduce to

The solution of the system of differential equations has the form

where the coefficients A _q depend upon the initial particle charge distribution through the recurrence law

The numerical model, Equations (1.a), (1b), (1c), and the analytical solution for pure photocharging, Equations (5.a), (5.b), (5.c), (5.d), were used to simulate the experiments on aerosol charging described in the next section. The values of the model parameters that appear in the ion-to-particle attachment coefficient rate η and in the particle photoionization rate α used in the simulations are listed in Table 1. The parameter K was determined by fitting to experimental data.

TABLE 1 Model parameters used in the simulations

Ion mobility	Z i	1.4· 10^−4 cm^2· V^−1· s^−1
Ion mass	m i	101 amu
Particle dielectric constant	ϵ r	10
Particle work function	Φ ∞	5.0 eV
Photon energy	hν	7.2 eV
Fowler law power	m	2

EXPERIMENTAL

The present work is aimed at demonstrating that photoionization is suitable for single unipolar charging of nanoparticles at high flow rates. For that purpose, a UV photocharger has been designed (Matter Engineering AG, Wohlen, Switzerland), which consists of a quartz (suprasil) tube of 16 mm in diameter and 400 mm in length surrounded by three lamps of Xe excimer that emits UV light with a wavelength λ of 172 nm (Heraeus Noblelight GmbH, Hanau, Germany).

Aerosol charging in the UV photoionizer has been studied in dependency of the particle size, the intensity of the radiation, and the irradiation time. The intensity I of the UV light emitted by the lamps varies from zero to an unknown maximum value, depending upon the intensity of the current supplied to the lamps i (≤ 0.5 mA). The exposure time of the aerosol to the UV radiation is approximated by the mean residence time of the aerosol in the device, defined as t _r = V/q, with V the volume of the quartz tube (80.43 cm³) and q the aerosol flow rate.

Two series of experiments were conducted at atmospheric pressure and room temperature (∼20°C). The dependence of the charging efficiency of the UV photoionizer on particle size was addressed in the first place. Figure 1 shows a layout of the set-up used in the experiments with monodisperse aerosols. A tin oxide powder (Acros Organics, New Jersey, USA) is evaporated in a three-zones tube furnace having an inner diameter of 15 cm and a length of 1.27 m (Carbolite, Hope Valley, UK) at intermediate temperatures (700–900°C) and carried in dry purified nitrogen at flow rates up to 100 l· min⁻¹ (Bronkhorst High-Tech BV, AK Ruurlo, The Netherlands). Downstream the furnace the gas becomes supersaturated with SnO vapor. This nucleates and forms small clusters which grow further by vapor condensation and Brownian coagulation. In that way, a polydisperse aerosol containing nanoparticles of a wide range of sizes is obtained.

FIG. 1 Set-up for charging of monodisperse aerosols with the UV photoionizer.

The aerosol is charged bipolarly in a radioactive ionizer containing a β -source of ⁸⁵Kr (10 mCi) and then it enters a DMA that operates at high aerosol flow rates (HF-DMA). The HF-DMA and the sheath gas recirculation line, which configuration is not displayed in Figure 1, have been described in a previous work (Hontañón and Kruis 2007). It was shown that the system formed by the ⁸⁵Kr charger and the HF-DMA is capable to deliver monodisperse nanoparticles in the size range of 5 to 25 nm with number concentrations between 10⁴ and 10⁵ cm⁻³ at flow rates as high as 100 l· min⁻¹.

The particles that leave the HF-DMA are singly charged, and they are partly neutralized in a radioactive ionizer similar to the one used to charge the polydisperse aerosol. At the outlet of the neutralizer, the remaining charged particles are removed from the aerosol in an electrostatic precipitator (ESP1). The size distribution and the number concentration of the uncharged particles are measured by using, respectively, a SMPS (TSI Inc., Minnesota, USA; TSI 3080) and a UCPC (TSI 3025). Then, the aerosol enters the UV photocharger. The total particle concentration N _p and the concentration of uncharged particles N ₀ at the exit of the UV photoionizer are measured with the UCPC by switching off and on, respectively, the voltage applied to an electrostatic precipitator (ESP2). The fractions of uncharged and charged particles in the aerosol are calculated as, respectively, f ₀ = N ₀/N _p and f _charged = 1− f ₀. Besides, the mobility distribution of the charged particles is measured with the Nano-DMA (TSI 3085) followed by the UCPC.

Secondly, the capability of the UV photoionizer for single unipolar charging of polydisperse aerosols in the nanometric range was assessed in comparison to the radioactive ⁸⁵Kr ionizer. The experimental set-up is depicted in Figure 2. Upstream the HF-DMA, the polydisperse aerosols are charged by alternatively the UV photoionizer or the ⁸⁵Kr ionizer. For given aerosol q and sheath Q flow rates, the voltage V applied to the HF-DMA is scanned from zero to the maximum attainable value, presently limited by the HV connection (± 20 kV); then, the particle size distribution and the particle number concentration of the aerosol selected with the HF-DMA are measured as a function of V by means of, respectively, the SMPS and the UCPC. If particles with multiple charges are present in the polydisperse aerosol, particles of different sizes and charge states having the same electrical mobility are classified together. In this case, the particle size distribution measured with the SMPS shows multiple peaks, corresponding to the various particle sizes.

FIG. 2 Set-up for charging of polydisperse aerosols with the UV photoionizer and with the ⁸⁵Kr ionizer.

Experiments were run at aerosol flow rates between 30 and 100 l· min⁻¹, with corresponding mean residence times in the UV photocharger of 0.16 to 0.048 s. A major difference between the two experimental series lies in the particle number concentration of the aerosols. The particle concentration of the polydisperse aerosols was more than three orders of magnitude higher than the particle concentration of the monodisperse aerosols.

RESULTS AND DISCUSSION

Monodisperse Aerosols

Uncharged monodisperse particles in the size range of 5 to 20 nm and number concentrations between 2· 10³ and 2· 10⁴ cm⁻³ were used in the first experimental series. We measured the fraction of charged particles f _charged and the mobility distribution of the particles at the outlet of the UV photoionizer. The experimental values of f _charged are plotted as a function of the particle size in Figure 3 for different irradiation levels and times. As expected, the charging efficiency diminishes with decreasing exposure time t _r , from 0.16 s (q = 30 l· min⁻¹) to 0.054 s (q = 90 l· min⁻¹), and when reducing the current intensity to the UV lamps i, from 0.5 mA to 0.1 mA.

FIG. 3 Charging efficiency of the UV photoionizer versus particle diameter under different irradiation conditions: t _r = 0.16 s (q = 30 l· min⁻¹), i = 0.5, 0.25, and 0.1 mA; and t _r = 0.054 s (q = 90 l· min⁻¹), i = 0.5 mA. Experimental values (symbols) and theoretical values (lines) calculated with Equations (5a)–(5d) for the indicated values of K.

The measurements with the Nano-DMA revealed the absence of negatively charged particles in the aerosols, indicating that diffusion charging played no role in the experiments. We attribute this to low ion production due to the low aerosol particle concentrations. On the other hand, diffusional losses in the UV photocharger reach at most 4.2%, for particles of 3 nm at a flow rate of 90 l· min⁻¹, as estimated by using a standard correlation for particle penetration in turbulent flow in smooth pipes (Baron and Willeke 1993). Moreover, in the experiments, the quartz tube was heated by absorption of UV light, so that the temperature of the tube was higher than the temperature of the aerosol passing through. Then, thermal forces develop radially which drive the particles toward the center of the tube.

Under the conditions above, i.e., negligible diffusion charging and particle losses, the analytical solution given by Equations (5a)–(5d) can be used to describe aerosol charging in the UV photoionizer. In Figure 3, the lines represent the analytical solutions obtained by best fitting to the experimental data, with the values of K displayed on the bottom right corner. For given irradiation conditions (t _r , I), a constant value of K results in good matching to the values of f _charged measured for particles between 8 and 20 nm. Thus, no dependence of the photoemission constant C on particle size was observed for particles of SnO in that size range. The uncertainty of the measurements increased largely with decreasing particle size, due to the low number concentrations and low charging efficiencies of such small particles; hence, the variation of C with particle size could not be discerned below 8 nm.

The values of K corresponding to a flow rate of 30 l· min⁻¹ in Figure 3 are plotted as a function of the current intensity i in Figure 4. The second order polynomial that matches the three data points is shown also. Taking into account that K = CI and, since the aerosol and the flow conditions were the same in the three experiments, C is constant, one concludes from Figure 4 that the intensity of the UV light emitted by the lamps does not vary linearly with the supplied current intensity. In addition, the polynomial predicts that I goes to zero at a value i ₀ of 0.0365 mA. This result is consistent with the observation of a threshold value of i ∼ 0.03 mA, below which charged particles were not detected at the outlet of the UV photoionizer in the experiments.

FIG. 4 The model parameter K vs. the intensity of the current supplied to the UV lamps. Experimental values (symbols) corresponding to t _r = 0.16 s (q = 30 l· min⁻¹), i = 0.5, 0.25, and 0.1 mA; and fitting polynomial (line).

On the other hand, we solved numerically the Equations (1a)–(1c) using the values of K determined previously and the values of the aerosol particle concentration N _p measured at the inlet of the UV photocharger. In accordance with the experimental observations, diffusion charging had no significance in simulations; the numerical solutions differed less than ± 2% from the analytical ones in all cases. Hence, we conclude that particle charging was dominated by direct photoionization.

The particle mobility distributions measured downstream the UV photocharger in the experiments at a flow rate of 30 l· min⁻¹ (t _r = 0.16 s) and current intensities of 0.5, 0.25, and 0.1 mA are displayed in Figure 5. The curves are normalized with respect to the peak value of singly charged particles (+1) at the maximum current intensity (0.5 mA). In addition, Figure 6 shows the particle charge distributions calculated with Equations (5a)–(5d) for the same conditions as in the experiments. Both the measurements and the calculations show the presence of particles with charge levels +2 and +3 in the experiments with current intensities of 0.5 and 0.25 mA. The model predicts also significant fractions of particles with charge levels +4 and +5 (Figure 6, top), which are not observed in the mobility curves corresponding to 0.5 and 0.25 mA (Figure 5). The mobility distributions corresponding to a current intensity of 0.1 mA show only singly charged particles up to 14 nm and some fractions of doubly charged particles above 14 nm, in reasonably good agreement with the theory (Figure 6, bottom-left).

FIG. 5 Particle mobility distributions measured at the outlet of the UV photocharger in experiments with monodisperse aerosols of the indicated particle diameters d _g . Irradiation conditions were t _r = 0.16 s (q = 30 l· min⁻¹), i = 0.5, 0.25, and 0.1 mA.

FIG. 6 Particle charge distributions of monodisperse aerosols calculated with the analytical solution for pure photocharging, Equations (5a)–(5d), for t _r = 0.16 s (q = 30 l· min⁻¹) and K = 6.5· 10³⁵, 3.75· 10³⁵ and 1.25· 10³⁵ J⁻¹· m⁻²· s⁻¹ (i = 0.5, 0.25, and 0.1 mA).

Particles with calculated fractions less than 5% could not be detected in the experiments. A plausible explanation is that, as the particle concentrations of the monodisperse aerosols were rather low, the concentrations of particles with such high number of charges in the aerosol at the outlet of the UV photoionizer were below the detection limit of the system formed of the Nano-DMA and the UCPC. The concentrations of particles larger than 14 nm with charge +5 were at most 2· 10² cm⁻³ in the simulations. In the experiments, particle losses in the sampling line and in the conduit through which the aerosol enters the Nano-DMA contributed to reduce further the concentration of highly charged particles. It is noticed that, to measure the particle mobility distribution downstream the UV photocharger, aerosol samples (1.5 l· min⁻¹) were taken from the main line and carried into the Nano-DMA.

On the other hand, the fraction of singly charged particles f _{+ 1} is a quite complex function of the particle size and the radiation intensity. In Figure 5 (top-right), particles of 14 nm attain about the same fractions of singly charged particles in the experiments with current intensities of 0.5 and 0.25 mA, while f _{+ 1} decreases by almost one-half in the experiment with a current intensity of 0.1 mA. Also in Figure 5, the fractions of singly charged particles of 17 nm (bottom-left) and of 18.5 nm (bottom-right) rise up when the current intensity is reduced from 0.5 to 0.25 mA and fall down when the current is lowered further to 0.1 mA; the decrease being more significant for particles of 17 nm. These experimental findings are corroborated by the theory, as can be seen in Figure 6 (bottom-right).

Figure 7 displays the particle mobility distributions measured at the outlet of the UV photocharger in experiments with reduced exposure time of 0.054 s (q = 90 l· min⁻¹) and the lamps operating at the maximum current intensity (0.5 mA). Much less multiple charging of particles is observed in these experiments, as compared to the experiments with an irradiation time three times longer and the same current intensity (Figure 5). The particle charge distributions predicted by the analytical model are plotted in Figure 8. The model predictions agree well with the measurements except for triply charged particles, which are not observed in the mobility curves in Figure 7; again, fractions of particles with charge +3 were below 5% in the calculations.

FIG. 7 Particle mobility distributions measured at the outlet of the UV photocharger in experiments with monodisperse aerosols of the indicated particle diameters d _g . Irradiation conditions were t _r = 0.054 s (q = 90 l· min⁻¹) and i = 0.5 mA.

FIG. 8 Particle charge distributions of monodisperse aerosols calculated with the analytical solution for pure photocharging, Equations (5a)–(5d), for t _r = 0.054 s (q = 90 l· min⁻¹) and K = 5.5· 10³⁵ J⁻¹· m⁻²· s⁻¹ (i = 0.5 mA).

Polydisperse Aerosols

The effect of the aerosol particle concentration, the radiation intensity, and the irradiation time on the charging of polydisperse aerosols in the UV photoionizer was investigated in the second experimental series. The results of three experiments with polydisperse aerosols, which conditions are summarized in Table 2, are discussed in this section.

TABLE 2 Description and conditions of the experiments with polydisperse aerosols

		Aerosol flow	Number concentration	Geometric mean	Geometric standard
Test	Description	rate q (l· min^−1)	N p (cm^−3)	diameter d g (nm)	deviation σ g
1	Reference case	30	2.1· 10^7	7.8	1.45
2	Effect of particle concentration	30	3.8· 10^7	9.4	1.51
3	Effect of irradiation time	90	3.2· 10^7	10.2	1.62

The intensity of the current supplied to the UV lamps was set to its maximum (0.5 mA) in the first experiment and we measured the particle size distribution of the aerosol that leaves the HF-DMA at different values of the applied voltage. The measurements revealed the presence of multiplets, i.e., particles with more than one charge, in the aerosols. Then, we reduced the current intensity until multiplets were not observed. Figure 9 (top) shows the normalized particle size distributions of the aerosols obtained with a voltage of +1.2 kV in the HF-DMA and with the UV lamps operating at current intensities of 0.5 and 0.1 mA. The latter led to a monodisperse aerosol with d _g ∼ 10 nm, σ _g ∼ 1.05 and N ∼ 5· 10⁴ cm⁻³.

FIG. 9 Particle size distributions measured at the outlet of the HF-DMA using the UV photoionizer to charge polydisperse aerosols with particle concentrations N _p of 2.1· 10⁷ cm⁻³ (top) and 3.8· 10⁷ cm⁻³ (bottom) at a flow rate q of 30 l· min⁻¹. Irradiation conditions were t _r = 0.054 s in both experiments, i = 0.5 and 0.1 mA (top), i = 0.1 and 0.04 mA (bottom).

The effect of the aerosol particle concentration was addressed in the second experiment, in which N _p enhanced in a factor of three. In this case, multiply charged particles were found in the aerosols at the outlet of the HF-DMA with the UV lamps operating at a current intensity of 0.1 mA. We attribute this to higher concentration of particles of sizes between 10 and 25 nm, attaining higher number of charges than particles less than 10 nm, the ones that contributed the most to the polydisperse aerosol in the first experiment. The current intensity was lowered down until multiplets disappeared. Figure 9 (bottom) displays the normalized particle size distributions of the aerosols corresponding to a voltage of +1.55 kV in the HF-DMA and the UV lamps working at current intensities of 0.1 and 0.04 mA. The latter resulted in a monodisperse aerosol with d _g ∼ 12 nm, σ _g ∼ 1.05 and N ∼ 5.5· 10⁴ cm⁻³.

The effect of the irradiation time was studied in the third experiment, in which the aerosol flow rate q increased in a factor of three. We set the intensity of the current to the UV lamps to 0.5 mA and, then, multiplets were not present in the aerosols at the exit of the HF-DMA. The reason is the reduced exposure time, with respect to the previous experiments. Well-defined monodisperse aerosols with particle sizes in the range of 5 to 20 nm were obtained by varying the voltage applied to the HF-DMA.

Experiments were performed also with the ⁸⁵Kr charger, the same polydisperse aerosols and the same flow rates as in the experiments with the UV photocharger. Particles less than 30 nm acquire at most one charge in a radioactive bipolar ionizer. Hence, in all the experiments with the ⁸⁵Kr charger the aerosols selected with the HF-DMA were monodisperse. The normalized particle size distributions of the monodisperse aerosols obtained with the ⁸⁵Kr charger and with the UV photocharger at a flow rate of 90 l· min⁻¹ are displayed in Figure 10. The peaks correspond to different values of the voltage applied to the HF-DMA, while the aerosol and sheath flow rates were kept constant and equal to, respectively, 90 and 1100 l· min⁻¹. A geometric standard deviation σ _g ∼ 1.05 was measured for particles of all sizes in the experiments with the two aerosol chargers.

FIG. 10 Particle size distributions of monodisperse aerosols measured at the outlet of the HF-DMA using the ⁸⁵Kr ionizer (top) and the UV photoionizer (bottom) to charge polydisperse aerosols with a particle concentration N _p of 3.2· 10⁷ cm⁻³ at a flow rate q of 90 l· min⁻¹. A geometric standard deviation σ _g ∼ 1.05 was obtained for particles of all sizes with the two aerosol chargers.

The particle number concentration N of the monodisperse aerosols, however, depends upon the single charging efficiency of the device used to charge the polydisperse aerosol upstream the HF-DMA. At the high flow rates and high particle concentrations of the polydisperse aerosols used in the experiments, charging conditions in the ⁸⁵Kr charger were unlikely stationary, so that the fractions of singly charged particles were lower than the theoretical ones (Wiedensohler 1988). On the other hand, aerosol charging by UV irradiation depends strongly upon the aerosol particle number concentration. As already mentioned, in the experiments with monodisperse aerosols particle concentrations were less than 2· 10⁴ cm⁻³ and particles were unipolarly positive charged. In the experiments with polydisperse aerosols, particle concentrations varied between 2· 10⁷ and 4· 10⁷ cm⁻³ and small fractions of negatively charged particles were found.

The single charging efficiency of the ⁸⁵Kr charger and of the UV photocharger was not measured. Instead, we measured the particle number concentration N of the monodisperse aerosol selected with the HF-DMA, as a function of the particle size and the particle polarity. The number concentrations of monodisperse particles obtained with the ⁸⁵Kr ionizer and with the UV photoionizer are compared to each other in Figure 11. As expected, slightly higher concentrations of negatively charged particles N _{− 1} are obtained with the ⁸⁵Kr ionizer, with respect to the positively charged ones N _{+ 1}. In the case of the UV charger, positively charged particles dominate clearly over negatively charged particles. As observed by comparison of the values of N _{± 1} measured at flow rates of 30 l· min⁻¹ (top) and 90 l· min⁻¹ (bottom), the concentration of singly charged particles diminishes with decreasing residence time, and this effect is more noticeable for the ⁸⁵Kr ionizer than for the UV photoionizer. In the latter, the reduced exposure time is compensated by increasing the intensity of the current supplied to the lamps.

FIG. 11 Comparison of particle number concentrations of monodisperse aerosols measured at the outlet of the HF-DMA using the ⁸⁵Kr ionizer and the UV photoionizer to charge polydisperse aerosols at flow rates q of 30 l· min⁻¹ (top) and 90 l· min⁻¹ (bottom).

The population balance equations, Equations (1a)–(1c), were solved numerically for a polydisperse aerosol having a flow rate q of 90 l· min⁻¹ and a log-normal particle size distribution with a particle number concentration N _p = 3.2· 10⁷ cm⁻³, a geometric mean diameter d _g = 10.2 nm and a geometric standard deviation σ _g = 1.62. We assumed a value of K of 5.5· 10³⁵ J⁻¹· m⁻²· s⁻¹, as derived from the experiments with monodisperse aerosols under the same irradiation conditions (t _r = 0.054 s, i = 0.5 mA). The model takes into account particle photocharging and particle discharging due to the attachment of negative ions to particles within the UV photoionizer, as well as diffusion charging in the pipe that connects the UV photoionizer and the HF-DMA and in the pipe from the aerosol inlet to the slit through which the aerosol enters the working region of the HF-DMA.

To a first approximation, we neglect ion and particle losses to the walls. The total fraction of charged particles and the particle charge distribution of the polydisperse aerosol at the exit of the UV photocharger calculated numerically are shown in Figure 12. The analytical solution for pure photocharging corresponding to the experiments with monodisperse aerosols is shown for comparison, as well. The polydisperse aerosol has a much higher particle number concentration, which results in largely enhanced ion production. Hence, differences between the polydisperse and the monodisperse cases in Figure 12 are due to diffusion charging. Figure 13 (top) displays the fractions of singly charged particles of both polarities f _{± 1} in the aerosol at the outlet of the UV photoionizer, at the aerosol entrance to the HF-DMA and at the aerosol slit. The ratio of singly negatively to singly positively charged particles N _{− 1}/N _{+ 1} at the same locations is plotted in Figure 13 (bottom) also, together with the values of N _{− 1}/N _{+ 1} estimated from the concentrations of singly charged monodisperse nanoparticles measured at the exit of the HF-DMA. Due to the attachment of negative ions to the particles, the fraction of singly positively charged particles f _{+ 1} reduces and the fraction of negatively charged particles f _{− 1} increases along the line. As observed in Figure 13 (bottom), the values of the ratio N _{− 1}/N _{+ 1} at the aerosol slit calculated numerically exceed largely the experimental values. Clearly the model overestimates diffusion charging, likely due to an excess of ions in the gas. We attribute this to neglecting ion losses in the calculation. Then, standard expressions of the ion depletion rate by diffusion to the walls in laminar (Maisels et al. 2003) and turbulent (Malet et al. 2000) flows in pipes were used for ω _i in Equation (1a) and a new calculation was performed. Diffusional losses had a minor impact on model predictions, with respect to the calculation with no ion losses.

FIG. 12 Comparison of the fraction of charged particles (top) and the particle charge distribution (bottom) at the outlet of the UV photoionizer calculated with the analytical solution for pure photocharging, Equations (5a)–(5d), for the experiments with monodisperse aerosols and with the numerical model including diffusion charging, Equations (1a)–(1c), for an experiment with a polydisperse aerosol. Irradiation conditions were t _r = 0.054 s (q = 90 l· min⁻¹) and i = 0.5 mA; ion and particle losses were not considered.

FIG. 13 Fractions of singly charged particles (top) and ratio of singly negatively to singly positively charged particles (bottom) in the aerosol at the outlet of the UV photoionizer, at the inlet of the HF-DMA, and at the aerosol slit calculated with the numerical model including diffusion charging, Equations (1a)–(1c), and experimental values of N _{− 1}/N_{+ 1} (open circles). Irradiation conditions were t _r = 0.054 s (q = 90 l· min⁻¹) and i = 0.5 mA; ion and particle losses were not considered in the calculation.

Mechanisms other than diffusion are not expected in the pipes downstream the UV photoionizer. However, the inside of the quartz tube might be not free of electric fields, since an alternating high voltage is supplied to the UV lamps around the tube (Matter et al. 1995). Assuming that the life-time of an ion τ _i (Z _i = 1.4 cm²· V⁻¹· s⁻¹) in presence of a steady uniform electric field E in a tube (φ = 16 mm) is given by τ _i = φ /2Z _i E, it is found that an electric field E _o ∼ 9 V· cm⁻¹ leads to a value of τ _i equal to the irradiation time of the aerosol in the experiment (t _r = 0.054 s). Hence, electric fields higher than E _o will deplete ions within the quartz tube. We carried out calculations with the ion depletion rate in the UV photoionizer w _i,UV as a free parameter and neglected ion losses in the pipes. Then, we determined the value of w _i,UV that best fit the values of the ratio N _{− 1}/N _{+ 1} measured in the experiments. A value of w _i,UV 77 s⁻¹ came out. A static uniform electric field of 38.5 V· cm⁻¹ would reduce the ion life-time to that value. Such weak electric fields could exist, even though one expects that the electric field within the quartz tube will be non-stationary and highly inhomogeneous.

Figure 14 displays the fractions of singly charged particles f _{± 1} (top) and the ratio of singly negatively to single positively charged particles N _{− 1}/N _{+ 1} (bottom) in the aerosol at the same locations as in Figure 13. As can be seen, the UV photoionizer behaves as a quasi-unipolar positive charger; fractions of singly negatively charged particles were less than 1% in the simulation. The estimated life-time of the ion in the UV photoionizer τ _i,UV is 0.013 s. For efficient single positive charging, τ _i,UV must be much shorter than the characteristic time for the attachment of negative ions to singly positively charged particles τ _{i, + 1} = (η _{+ 1} N _{+ 1})⁻¹. For the particle sizes of the aerosol used in the experiment (3–30 nm), η _{+ 1}varies between 10⁻⁶ and 2· 10⁻⁶ cm³· s⁻¹ and, assuming a particle concentration N _{+ 1} of 10⁷ cm⁻³, the ion depletion time τ _{i, + 1} lies in the range of 0.1 to 0.05 s. In the calculation without ion losses, τ _i,UV = t _r = 0.054 s ≈ τ _{i, + 1}, and diffusion charging was noticeable. In the calculation with enhanced ion depletion in the irradiation region, τ _i,UV = 0.013 s ≪ τ _{i, + 1}, and diffusion charging had a minor effect, as compared to the first calculation.

FIG. 14 Fractions of singly charged nanoparticles f _{± 1} (top) and ratio of singly negatively to singly positively charged particles N _{− 1}/N _{+ 1} (bottom) in the aerosol at the outlet of the UV photoionizer, at the inlet of the HF-DMA, and at the aerosol slit calculated with the numerical model including diffusion charging, Equations (1a)–(1c), and experimental values of N _{− 1}/N _{+ 1} (open circles). Irradiation conditions were t _r = 0.054 s (q = 90 l· min⁻¹) and i = 0.5 mA; an ion depletion rate w _i,UV of 77 s⁻¹ was assumed in the calculation.

Therefore, for efficient single unipolar charging of polydisperse aerosols with nanoparticles smaller than 30 nm and number concentrations in the order of 10⁷ cm⁻³ by UV photocharging, the life-time of the ions in the aerosol must be less than 10 ms. In the UV photocharger used in this work, the irradiation region was not shielded, so that electric fields developed which contributed to remove ions from the aerosol.

CONCLUSIONS

This work proves the feasibility of single unipolar charging of aerosols by UV photoionization and its applicability to the DMA-based production of monodisperse nanoparticles at flow rates up to 100 l· min⁻¹. Thus, UV irradiation is a promising aerosol charging technique for application to mass production of monodisperse nanoparticles in gas phase by means of electrical mobility analysis.

UV photocharging is rather sensitive to the composition, particle size distribution, and particle number concentration of the aerosol. The energy of the UV radiation must exceed the work function of the aerosol material. For the production of monodisperse particles, a UV unipolar photocharger is an optimal choice. For that purpose, diffusion charging, which leads to a decrease in the unipolar charging efficiency, can be inhibited by installing an ion trap in the irradiation region. On the other hand, UV photochargers should be equipped with the possibility to modify the intensity of the radiation. Depending on the aerosol particle number concentration, the flow rate and the particle size to be selected with the DMA, the radiation intensity need to be optimized to minimize multiple charging and to maximize the single charging efficiency.

A major drawback of the current models on aerosol photocharging is an empirical parameter K, which is the product of the photoemission constant and the radiation intensity, both unknown quantities difficult to measure in practice. A method has been developed to estimate the values of K from experimental data on photocharging of monodisperse aerosols.

Acknowledgments

This work was supported by the Deutsche Forschungsgemeinschaft (DFG) in the framework of the Collaborative Research Centre on “Nanoparticles from the gas phase: formation, structure and properties” (SFB 445). Dr. Esther Hontañón was supported by Ministerio de Educación y Ciencia (MEC) of Spain.