The use of the pulse height analyser ultrafine condensation particle counter (PHA-UCPC) technique applied to sizing of nucleation mode particles of differing chemical composition

Abstract

A modified white-light pulse-height analyser (PHA) TSI 3025 ultra-fine condensation particle counter (UCPC) is often used to provide fast response of aerosol size distributions between 3 and $\text{10}\text{nm}$ since there is a monotonic link between initial aerosol size and nucleated droplet final size. The use of the PHA-UCPC for sizing nucleation mode particles depends on the droplet nucleation in the condenser chamber being somewhat independent of particle composition. Laboratory characterization of the PHA-UCPC for a range of chemical compositions, thought to be involved in atmospheric aerosol nucleation and growth, are presented here. Ammonium sulphate, pinic acid, cis-pinonic acid, malonic acid and an iodine oxide were studied and their PHA-UCPC calibration kernels are presented. It was found that all species possessed significantly different PHA responses. The results suggest that, unless the nano-particle chemical composition is known, then the PHA-UCPC cannot be used for measurements of aerosol size distributions. However, the PHA-UCPC, if used in parallel with mobility size distribution measurements, can help elucidate nano-particle chemical composition. Using the combination of mobility size distributions and the PHA-UCPC response during a nucleation and growth event over the boreal forest indicates that new particle formation, in this region, is driven by condensation of organic vapours.

Introduction

Bursts of new atmospheric aerosol particles have been encountered in non-perturbed, or background, air in recent years. Regular formation-bursts of 3–$\text{5}\text{nm}$ particles are observed, particularly in Spring, over the boreal forest of Finland (Mäkelä et al., 1997; Kulmala et al., 2001) along with forests in southern Europe (Kavouras, Mihalopoulos, & Stephanou, 1998). These bursts are normally followed by growth periods where the new particles grow, primarily by condensation, into accumulation mode sizes $\text{(d>100}\text{nm}\text{)}$. Similar events have also been observed in rural Germany by Birmili, Wiedensohler, Plass-Dülmer, and Berresheim (2000). Nucleation bursts have also been observed in the coastal environment and have been studied in detail by O'Dowd et al. (1999 and O'Dowd, Hämeri et al. (2002). Less frequent, but still observable, are bursts in the clean marine boundary layer (e.g. Clarke et al., 1998). During nucleation bursts, the size of the new particles can be determined by measuring the total particle concentration using condensation particle counters (CPCs) possessing a (50% efficiency) lower cut-off detection size of 3, 5 or $\text{10}\text{nm}$ or by using a differential/scanning mobility particle sizer (D/SMPS) (Hämeri, O'Dowd, & Hoell, 2002). While running CPCs in parallel with different cut-off sizes leads to high time resolution required during some dynamic nucleation events, the size resolution is somewhat crude. On the other hand, while D/SMPS measurements can provide high size resolution data, the temporal resolution is poor with a typical resolution of between 2 and $\text{20}\text{min}$ depending on the chosen configuration of the system.

In an attempt to improve the rapid sizing of nucleation mode particles, Saros, Weber, Marti, and McMurry (1996) and Weber et al. (1997), Weber, Stolzenburg, Pandis, and McMurry (1998) modified a TSI 3025 CPC to extract continuous size distributions of nucleation mode particles at a temporal frequency of the order of seconds. The principal behind this so-called PHA-UCPC is that when particles larger than $\text{10}\text{nm}$ are activated in the butanol condenser (cloud) chamber, the activated drops, during the growth process, converge to one final size by the time they exit the condensing chamber and enter the detection optical chamber. As a result of all droplets converging to one final size, they all produce one characteristic pulse from the scattered detector light. For particles between 3 and $\text{10}\text{nm}$, due to the greater Kelvin effect at these sizes, their growth time in the chamber is insufficient for them to “catch-up” with the bigger droplets. Consequently, each initial particle below $\text{10}\text{nm}$ produces a unique pulse height which can be monotonically linked to the initial nucleus size. To achieve this, however, the laser light in the UCPC must be replaced with a white light source to overcome the Mie multiple scattering response effect (Marti, Weber, Saros, & McMurry, 1996; Saros et al., 1996). The deployment of the PHA-UCPC has the potential to overcome the difficulties of conducting fast-response measurements of nucleation mode size spectral evolution over short time-scales.

Greater difficulties are associated with the identification of the chemical composition of these nanometer particles due to the difficulties in separating this mode from the existing aerosol population, while maintaining sufficient mass to conduct chemical analysis. Conventional filter or impactor techniques simply cannot adequately separate particles smaller than $\text{10}\text{nm}$ from the pre-existing aerosol and, consequently, any chemical analysis conducted will be significantly influenced by particles larger than $\text{10}\text{nm}$. Electrical precipitation can efficiently separate the nucleation mode below $\text{10}\text{nm}$ from the pre-existing aerosol, however, given the charging efficiency of the order of 4% for $\text{10}\text{nm}$ and 1% for $\text{3}\text{nm}$ particles, there are only a few particles separated for either bulk analysis or single particle analysis. It should be noted, however, that in the coastal environment, where nucleation mode concentrations exceeding $\text{10}{}^6\text{cm}{}^{\mathrm{-3}}$ are often encountered, electrical separation was found to be successful and sufficient particles in the size range of 6–$\text{8}\text{nm}$ were sampled for subsequent high resolution Transmission Electron Microscopy where iodine and sulphate could be identified as the primary particle composition (Mäkelä et al., 2002). While this technique proved successful in the coastal environment, for other regions of natural particle formation (remote ocean, free troposphere or over forest), it cannot yield any useful results.

A more dynamic method to study the evolution of nucleation mode composition, albeit indirectly, has been deployed by Hämeri et al. (2001) where the hygroscopic properties of nucleation mode particles are examined. This technique can typically distinguish between species which are soluble or non-soluble in water. In the coastal environment, it was found that $\text{8}\text{nm}$ particles were rather non-soluble in water and were thought to comprise an iodine oxide composition (Väkevä, Hämeri, & Aalto, 2002; O'Dowd, Hämeri et al., 2002), while in the forest environment, $\text{10}\text{nm}$ particles were found to have varying degrees of solubility which possessed a diurnal cycle. In the latter, it was concluded that a significant fraction of the nucleation mode chemical composition comprised water-soluble organics.

While the aforementioned studies represent major breakthroughs in identification of recently-formed particles composition, they still do not necessarily elucidate the new particle formation processes since the composition of $\text{10}\text{nm}$ particles can differ significantly from younger 1–$\text{3}\text{nm}$ particles. Consequently, the identification of particle composition, particularly in the 3–$\text{6}\text{nm}$ size range (which can be defined as detectable particles rather than new particles of $\text{1}\text{nm}$) is crucially important to elucidating which species actually produce particles in the natural atmosphere.

In this study, we characterize the response function of a PHA-UCPC to a number of a sub–$\text{10}\text{nm}$ aerosol species thought to be involved in new particle production and we illustrate, following O'Dowd, Aalto, Hämeri, Kulmala, and Hoffmann (2002), that a combination of mobility based (DMPS) and condensation growth (PHA-UCPC) measurements during nucleation bursts can be used to elucidate 3–$\text{6}\text{nm}$ nucleation mode aerosol composition, although, by the nature of the technique, at the expense of retrieving nucleation mode size spectra with the PHA-UCPC.