The formation of H2SO4/H2O particles has been investigated in a laboratory study using a flow tube at atmospheric pressure. H2SO4 was produced in situ applying different formation pathways. Experiments performed in the absence of organics as well as those in the presence of selected organics with low concentration yielded nearly the same results indicating that organics are not implicitly needed in the formation process of atmospheric H2SO4/H2O particles. The number of newly formed particles increased with increasing humidity in the range r.h. = 11–60%. Measurements of particle size distributions at r.h. = 42% revealed a mean particle diameter of ∼2 nm for an H2SO4 concentration of ∼107 molecule cm−3 and for a residence time of 290 sec. From the measured fraction of positively and negatively charged particles it was concluded that ion-mediated nucleation was negligible under the chosen experimental conditions.
 Atmospheric particles have a strong impact on the Earth's radiation budget due to their radiative properties and the fact that they can act as condensation nuclei for clouds [Charlson and Heintzenberg, 1995]. Field measurements at ground level show atmospheric nucleation events for H2SO4 concentrations of ∼107 molecule cm−3 [Weber et al., 1999; Fiedler et al., 2005]. Despite intensive research activities in the last decade, the mechanism leading to new particles has not been unambiguously revealed yet [Kulmala, 2003]. Currently, mechanistic approaches involve binary nucleation H2SO4-H2O [Kulmala et al., 1998], ternary nucleation H2SO4-H2O-NH3 [Korhonen et al., 1999], ion-mediated nucleation [Lovejoy et al., 2004], nucleation processes supported by organics [Zhang et al., 2004] or a theory of cluster activation [Kulmala et al., 2006]. Results of appropriate models are partly controversial, for example, in the case of ternary H2SO4-H2O-NH3 nucleation the model of Napari et al.  suggests nucleation for atmospheric conditions while Yu [2006a] shows that ternary nucleation should be of less importance. For ion-mediated nucleation, the model of Lovejoy et al.  results in very small nucleation rates while that of Yu [2006b] yields significant nucleation rates.
 In a previous investigation from this laboratory, experimental evidence for the formation of new particles in the system H2SO4/H2O under near-atmospheric conditions with H2SO4 concentrations of ∼107 molecule cm−3 was found [Berndt et al., 2005]. H2SO4 was produced in situ via the reaction of OH radicals with SO2 in the presence of water vapor. The photolysis of O3 served as the OH radical source. Organics were added for determining the OH radical concentration (and consequently the H2SO4 concentration). Particle formation was also observed for in situ produced H2SO4 concentrations of ∼107 molecule cm−3 using the ozonolysis of alkenes as a “dark” OH radical source [Berndt et al., 2004].
 Similar observations are reported from a study in the 590 m3 Calspan chamber [Verheggen, 2004]. Here, new particle formation took place for in situ produced H2SO4 concentrations of a few 107–108 molecule cm−3 being determined by means of chemical ionization mass spectrometry measurements or with the help of calculations using a kinetic model.
 Subject of this study is to investigate whether the low threshold H2SO4 concentration for nucleation of ∼107 molecule cm−3 for in situ produced H2SO4 arises from any contributions of organic substances being present in the experiments. A further point of interest is the charge distribution of the particles formed. With the knowledge of that the importance of ion-mediated nucleation in the system can be evaluated.
2. Experimental Methods
 The experiments have been performed in the IfT-LFT (Institute for Tropospheric Research – Laminar Flow Tube; i.d. 8 cm; length 505 cm) at atmospheric pressure and 293 ± 0.5 K. The flow tube consists of three sections: a first section (56 cm) includes an inlet system for gas entrance, a middle section (344 cm) with 8 UV lamps for a homogeneous irradiation if needed, and a non-irradiated end section (105 cm) for the sampling outlets. The concentrations of the organic gas-phase species have been measured by means of on-line GC-FID analysis (HP 5890). For ozone concentration measurements, an ozone analyzer (Thermo Environmental Instruments 49C) as well as long-path UV spectroscopy (PerkinElmer Lambda 800; path length 512 cm) have been utilized. Synthetic air (99.9999999%, Linde 99.999% and further purification with Gate Keeper, AERONEX) served as the carrier gas. Ozone was generated outside the IfT-LFT by means of a flow-through photolysis reactor. The total gas flow was set at 3.6 liter min−1 resulting in a residence time in the middle section of the IfT-LFT of 290 sec.
 Particle size distributions (dp > 2 nm) were determined using a differential mobility particle sizer (DMPS) consisting of a short Vienna-type differential mobility analyzer (DMA) and an ultrafine particle counter (UCPC, TSI 3025). Charged particles were measured after removal of the Kr-85 charger from the DMPS system. Changing the DMA polarity allowed counting of positively and negatively charged particles. In addition, the total number of particles was determined by a second UCPC (TSI 3025) directly at the outlet of the IfT-LFT.
2.1. Determination of H2SO4 Concentration
 Like in the atmosphere [Finlayson-Pitts and Pitts, 2000], H2SO4 was produced in situ via the reaction of OH radicals with SO2 in the presence of water vapor. In most cases, photolysis of O3 with the subsequent reaction of formed O(1D) with water served as source of OH radicals. Furan (organic) was added for OH radical titration. The measured furan decay allowed the calculation of the OH radical profile in the IfT-LFT and, consequently, the H2SO4 concentration [Berndt et al., 2005].
Simultaneous measurements of the furan and the O3 decay yielded a ratio “reacted O3”/“reacted furan” = 0.5 ± 0.2 being in line with the simple reaction pattern given above. Per “effective” photolyzed O3 molecule two OH radicals were formed. (Note, for OH radical balance, pathway (6) is negligible using furan in excess over SO2.) Therefore, pathways (1)–(4) can be combined resulting in an “effective” O3 photolysis step.
Assuming a constant light intensity throughout the irradiated section an “effective” rate coefficient k1a can be determined from the observed O3 decay. So, substituting furan by CO, that is, removing the organics from the system, the H2SO4 concentrations in the tube were calculated by numerical integration of the resulting differential equations of pathways (1a) and (5)–(7). Necessary rate coefficients were taken from literature [Finlayson-Pitts and Pitts, 2000; Hanson and Eisele, 2000]. Wall loss reactions excepting pathway (7) were of less importance under the chosen conditions.
 In addition, in a dark reaction, the needed OH radicals were produced via ozonolysis of t-butene. The procedure of H2SO4 determination was similar to that given elsewhere [Berndt et al., 2004].
 In this study, the experiments have been conducted in synthetic air. A carrier gas composition of 1 vol% O2 in N2 was chosen in earlier investigations [Berndt et al., 2005]. To make all findings comparable, runs with O3 photolysis in the presence of furan or ozonolysis of t-butene [Berndt et al., 2005, 2004] were repeated in this work using synthetic air. Figure 1 shows the experimental results for the different approaches for OH radical generation, O3 photolysis in the presence (furan) or absence (CO) of organics and ozonolysis of t-butene (dark reaction). For comparison, in Figure 1 results from an earlier set of runs (O3 photolysis in the presence of furan) using 1 vol% O2 in N2 as the carrier gas are added. The findings reveal that the change of the O2 concentration (1 vol% or 20 vol%) does not influence the particle formation. All four series consistently demonstrate that new particle formation starts for in situ produced H2SO4 concentrations of ∼107 molecule cm−3. Therefore, it has to be concluded that organics are not implicitly needed for the process of particle formation under these conditions. Consequently, the different threshold H2SO4 concentrations of ∼107 molecule cm−3 (in situ formation) and ∼1010 molecule cm−3 (from bulk sample) cannot be explained by a particle formation mechanism mediated by organics.
 The stated H2SO4 concentrations are calculated end concentrations in the flow tube where a nearly constant H2SO4 concentration builds up after few seconds. Uncertainties from the needed measurements (reacted furan or O3) as well as errors of the rate coefficients for pathways (1a) and (5)–(7) can cause an uncertainty of calculation H2SO4 concentrations of a factor of ∼2. Differences between the four series among each other can be explained in this way. The only H2SO4 sink considered in the calculations is the wall loss via pathway (7). Modeling shows a continuous increase of H2SO4 with time resulting in a ∼5-fold final concentration neglecting this sink. Therefore, the maximum uncertainty of H2SO4 concentrations should be not larger than a factor of 5. (An example of H2SO4 profiles in the tube is given in auxiliary material).
 The application of long-path UV spectroscopy improved significantly the reliability of O3 measurements being the crucial point for H2SO4 determination in the cases where OH was produced via photolysis of O3 (presence of CO) and via ozonolysis of t-butene. This is the reason for the shift down of calculated H2SO4 concentration by a factor of ∼2 compared to the results presented in a former report [Berndt et al., 2004].
 In Figure 2 experimental data for different r.h. are depicted as resulting from O3 photolysis runs in the absence of organics (presence of CO). The measurements show a distinct impact of the water vapor content on the number of newly formed particles. For example, for an H2SO4 concentration of 107 molecule cm−3 the particle number increased in the order <1, 20, 250, and 800 cm−3 for r.h. = 11, 22, 42, and 60%, respectively. The effect of increasing particle number with increasing r.h. is more pronounced in the range of lower r.h. From a mechanistic point of view, the strong r.h.-dependence of the number of newly formed particles indicates that water vapor has to be involved in the process of particle formation and/or particle growth.
 Experiments with enhanced H2SO4 concentrations allowed the determination of particle size distributions. Figure 3 shows DMPS measurements for three different H2SO4 concentrations at r.h. = 42%. Even for the highest H2SO4 concentrations, the mean particle diameter was below 3 nm. A particle diameter of ∼2 nm for an H2SO4 concentration of 107 molecule cm−3 can be roughly estimated from the observed trend of decreasing particle diameter with decreasing H2SO4 concentration, cf. Figure 3.
 Surprisingly, particle counting was possible for diameters down to ∼2 nm (stated TSI 3025 cut-off: 3 nm). In order to investigate this fact, simultaneous measurements were undertaken using an ultrafine water-based counter (TSI 3786) with an improved cut-off of 2.5 nm for comparison. Both counters (TSI 3025 and TSI 3786) yielded the same results performing either size distribution measurements with the DMPS system or integral measurements of the total particle number. Furthermore, in a few runs the DMPS system was replaced by a Nano-SMPS system (TSI 3936-N86 with TSI 3786). Resulting size distributions were in reasonable agreement with those from the DMPS system. These findings suggest that in the case of H2SO4 particles, efficient counting seems to be possible for a particle diameter of ∼2 nm and probably below. Obviously, the counting efficiency for small particles is dependent on their chemical composition. In the case of H2SO4 particles, the performance of CPC's is excellent, cf. for example. [Madelaine and Metayer, 1980].
 Integration of the measured particle size distributions allowed the determination of total particle volume. Assuming that the particles consist of H2SO4 only (with a density of 1.85 g cm−3), the upper limit of the yields of particulate H2SO4 regarding total H2SO4 can be given. According to this, a yield of 0.9, 5, and 9% was found for H2SO4 vapor concentrations of 2.0 · 107, 3.6 · 107, and 4.5 · 107 molecule cm−3, respectively, cf. Figure 3. Because of the lack of any information concerning the water content of these particles, a more accurate determination of particulate H2SO4 is not possible at this time.
 It should be noted that at H2SO4 concentrations in the order of 107 molecule cm−3 the observed particle growth cannot be due to commonly used condensation laws of H2SO4 onto nucleated particles. However, if classical nucleation theory and the underlying thermodynamics are not able to properly describe the nucleation itself, it seems to be impossible applying these thermodynamics to model the growth of the freshly nucleated particles. However, if kinetically limited growth is assumed, H2SO4 concentrations in the order of 107 molecule cm−3 are sufficient to explain the particle growth observed. This is supported by model calculations. Note, “in situ produced H2SO4” means that in the course of nucleation and particle growth O3, HO2 and H2O2 are present in the reaction gas being in concentrations close to atmospheric levels. For HO2-H2SO4 a very stable adduct is proposed [Miller and Francisco, 2001]. Such adducts can support particle growth. Because no experimental evidence exists so far, this process remains highly speculative at the moment.
 Measurements of particle size distribution together with the corresponding distributions of negatively and positively charged particles were performed for H2SO4 concentrations of (2.8–7.3) · 107 molecule cm−3 and for r.h. = 42%. (An example is given in auxiliary material.) As a result of integration, the number of negatively charged particles relative to the total number was in the range 0.0003–0.0008. The number of negatively charged particles exceeded the number of positively charged particles in all cases. In order to investigate the possible role of UV-light induced ion formation, similar experiments have been conducted using ozonolysis of t-butene for OH radical generation (dark reaction). As the result, there were no significant differences between photolysis and dark experiments. That points at terrestrial or cosmic ray as the possible sources for ions in the flow tube [Rosen et al., 1985]. After all, the measurement of the fraction of charged particle shows clearly that more than 99% of the newly formed particles are uncharged. Therefore, ion-mediated nucleation is of minor importance under the experimental conditions chosen here.
 The formation of H2SO4/H2O particles has been studied under near-atmospheric conditions using an atmospheric pressure flow tube. H2SO4 was produced in situ via the reaction of OH radicals with SO2 in the presence of water vapor. Experiments performed in the absence of organics as well as those in the presence of selected organics with low concentration yielded approximately the same results. This indicates that organics are not necessarily needed in the formation process of atmospheric H2SO4/H2O particles. On the other hand, this finding does not rule out that other selected organics might support or trigger atmospheric new particle formation. The observed r.h.-dependence of the number of newly formed particles indicates that water vapor has to be involved in the process of particle formation and/or particle growth. From the measurements of particle size distributions a mean particle diameter of ∼2 nm can be deduced for an H2SO4 concentration of ∼107 molecule cm−3 (r.h. = 42%, residence time of 290 sec). The measured fractions of positively and negatively charged particles indicated that ion-mediated nucleation was negligible in the experiments.
 Finally, the results from this study show that the low threshold concentration for nucleation of ∼107 molecule cm−3 in the case of in situ produced H2SO4 (from bulk sample ∼1010 molecule cm−3 of H2SO4) cannot be explained by a contribution of organics supporting particle formation or by ion-mediated nucleation. Probably, in the course of the SO2 oxidation (SO2 → SO3 → H2SO4) other species than the expected H2SO4(-hydrates) are formed being responsible for particle formation. That is highly speculative at present and much more work is needed for clarifying the nucleation mechanism.
 The authors thank O. Bischof and T. Krinke from TSI for reference measurements with an ultrafine water-based counter (TSI 3786) and a Nano-SMPS system (TSI 3936-N86 with TSI 3786) as well as K. Pielok and H. Macholeth for technical support.