A PM 1 . 0 / 2 . 5 / 10 Trichotomous Virtual Impactor Based Sampler : Design and Applied to Arid Southwest Aerosols Part II : Application to Arid Southwest Aerosols †

A PM1.0/2.5/10 Trichotomous sampler has been developed to determine if the particles in the saddle point between the coarse and fine particle modes (specifically the 1.0 μm to 2.5 μm size range) are primarily coarse or fine mode particles. The sampler consists of a standard high volume sampler with two high volume virtual impactors (one with a cut size of 2.5 μm and the other with a cut size of 1.0 μm) inserted between the PM10 inlet and the 8 × 10 inch after filter. Filters in this sampler were analyzed with ion chromatography (IC) to determine SO4 concentrations, representing a specie primarily found in the fine mode aerosol and proton-induced X-ray emission (PIXE) for determining concentrations of Si, S, Ca and Fe, representing species normally found in coarse mode aerosols. Application of this sampler to Phoenix, AZ, representing an arid region, showed that particles in the saddle point consisted of about 75% of particles from the coarse mode and about 25% from the fine mode.


Introduction
In the mid-1990s the United States Environmental Protection Agency (EPA) was reviewing the National Ambient Air Quality Standards (NAAQS) for particulate matter (PM).The agency was devising a new Federal Reference Method (FRM) and set a new fine particle standard at either PM 1 or PM 2.5 to replace the then current 24-hour PM 10 standard of 150 micrograms per cubic meter (μg/m 3 ).When the fine particle standard was being established, there was much discussion amongst experts as to whether the standard should be set at 1.0 μm or 2.5 μm.The problem arose from the nature of the particle size distributions in atmospheric aerosols.Whitby (1978) discovered that atmospheric particle size distributions were trimodal in nature, with fresh combustion particles in the smallest mode, aged combustion particles in the intermediate mode (fine particle mode) and mechanically generated particles in the largest mode (coarse particle mode).These fine and coarse particles have different chemical and physical characteristics and come from different sources (John, 2011).Dr. J.P. Lodge stated in 1995 in the Journal of the AWMA in a "Critical Review Discussion: Measurement Methods to Determine Compliance with Ambient Air Quality Standards for Suspended Particles" (Watson et. al., 1995) that: "I had a lengthy discussion with the late K.T. Whitby over the optimum upper boundary for fine particles, which he held to be "... 1.8 μm, or, if you want a purer sample of fine particles, 1 μm."If fine particles are the cause of health effects, and if routine monitoring data are used in epidemiological studies, as Dr. Chow anticipates, we must advocate the use of PM 1.0 to clarify the composition of the population of the fine particle modes." This statement would indicate that an ideal particle cut size would be one below which almost all the fine particles are contained, but very few coarse particles are present.Particles in this size range consist of particles from the lower tail of the coarse particle mode and particles from the upper tail of the fine particle mode (see Fig. 1).
Of specific interest in this study was the atmospheric particulate matter of the arid southwest USA.The city chosen to study was Phoenix, Arizona from May to October, 1995.In this area of the country the coarse particle mode can be much larger than the fine particle mode and, thus, may dominate the particulate matter in the saddle region between the fine and coarse particle modes.
The concern was that the large particle mode lower tail, consisting of fugitive and windblown dust, could dominate the particles in the 1 to 2.5 μm size range, a test program was initiated to collect particulate matter in this size range and determine the fraction of the particulate matter that may be coarse mode aerosol.
In part I of this paper, published in the 2012 issue of KONA (Marple et.al., 2012), a PM 1.0/2.5/10trichotomous sampler (see Fig. 2) was developed to specifically answer the question of what fraction of particles in the 1.0 μm to 2.5 μm size range are part of the fine particle mode and what fraction are from the coarse particle mode, is described.
By using nine 47 mm filters at various locations in the trichotomous sampler, concentration and composition of particles could be obtained for particulate matter in the ranges of PM 10 , PM 2.5-10 , PM 2.5 , PM 1-2.5 , and PM 1 .From data analysis of these filters, the fraction of the particulate matter in the 1.0 μm to 2.5 μm size range that are fine and coarse mode particles can be determined.See Part I for further details of this sampler.

Mass and specie analysis techniques
The net mass gains on the filters were determined gravimetrically by weighing them before and after sampling.The weighings were conducted in a controlled environment where the temperature was maintained at 24 ± 3°C and the relative humidity at 30 ± 10% to minimize loss of soluble and volatile compounds.As a quality control measure, approximately one out of ten filters was later reweighed.
Several methods were available to determine elemental compositions of the PM collected on the filters.For this study, proton-induced X-ray emission (PIXE) and photon-induced x-ray fluorescence (XRF) methods were used.Both of these methods are commonly used due to their nondestructive multi-element capabilities and relatively good sensitivities.Aerosol ions (i.e., the water-soluble portion of the suspended particles in the atmosphere) were quantified using ion chromatography (IC).Using the IC method, the extracted sample passes through an ion-exchange column and the separated ions are individually analyzed by an electroconductivity detector.
Volatile aerosol materials such as organics and nitrates were not included in this study.Although a large portion of the fine aerosol often consists of volatile organics or nitrates, little of the remaining (weighable) PM collected on the filters was expected to be volatile, since the filters were exposed to the high summer temperatures of Phoenix (often exceeding 40°C during the day).

Mass concentrations
PM mass concentrations (μg/m 3 ) were determined by dividing the net weight gain (μg) on a filter by the total air volume (m 3 ) that passed through it during the 24-hour sampling period.Although a small decrease in flowrate can be expected at the end of sampling period due to a slightly higher pressure drop caused by particle loading on the filters, no flowrate correction was deemed necessary for the purposes of this study.Mass concentrations were, however, divided by an appropriate "flow factor" to account for the fact that the samplers did not run for exactly 24 hours in most cases.The general equation used for mass calculations is as follows: Where PM X = PM mass concentration for size fraction × (μg/m 3 ), Δm = mass gain on filter (μg), V = flow volume through filter in 24 hours (m 3 ) and f = flow factor (fraction of 24-hour day that sampler was operated).Thus, the following equations were then used to calculate mass concentrations for various PM size fractions: In equations ( 5) and ( 6), the terms {Δm(PM 2.5-10 )} and {ΣΔm(PM 1-2.5 )} are the masses collected on the minor flow filters of the 2.5 μm and 1.0 μm HVVIs, respectively.The Σ sign in equation ( 6) indicates summing the mass collected on both filters being used to collect minor flow particles in the 1.0 μm HVVI.As described in Part I of this paper, particles collected in the minor flow of the HVVIs contain background particles (particles of a size less than the cutsize of the HVVI) and the mass of these background particles must be subtracted from the mass of the particles in the minor flow to determine the true mass of particles in the minor flow that are larger than the cutsize of the HVVI.In equations ( 5) and ( 6), this background mass is represented by the terms {Δm(PM 2.5 )/10} and {ΣΔm(PM 1 )} in the 2.5 μm and 1.0 μm HVVIs, respectively.The factor of 10 in the denominator of equation ( 5) is due to the flowrate of the filter collecting the major flow particulate matter of the 2.5 μm HVVI is 2 cfm and the flowrate of the filter collecting particulate matter in the minor flow is 0.2 cfm.
Using equations ( 2) to (6), results of the 6-month sampling program in Phoenix are summarized as mass concentrations in Table 1 and as percent of PM 10 in Table 2. Average PM 1 , PM 2.5 and PM 10 concentrations for the Phoenix area are 5.9, 8.6, and 32.8 μg/m 3 , respectively.In terms of percentages, PM 10 was made of 69% PM 2.5-10 , 26% PM 2.5 (< 2.5 μm), 8% PM 1-2.5 , and 18% PM 1 (< 1 μm).It is also interesting to note that average % PM 1-2.5 appears to be nearly constant (at about 8%) for all six months.A comparison of measured PM 10 and summed PM 10 (PM 2.5-10 + PM 1-2.5 + PM 1 ) can also be seen in Tables 1 and 2. In general, summed PM 10 is slightly less than measured PM 10 , most likely due to particle loss in the HVVIs.

Intermodal size range (PM 1-2.5 ) calculations
Atmospheric aerosol is formed by two basic processes: the dispersion process, which is the breakup of solid and liquid particles; and the condensation/reaction process, which represents particle formation by coming together of gases and smaller particles (including surface reactions).Aerosol formed by the dispersion mechanism is generally of larger size (normally > 1 μm in diameter) and will be referred to as "coarse" mode aerosol (and the particles as coarse particles) (ref).On the other hand, aerosol formed by the condensation/reaction mechanism is usually of smaller size (normally < 2.5 μm in diameter) and will be referred to as "fine" mode aerosol (and the particles as fine particles).All diameters are of aerodynamic particle size (i.e., a unit density spherical particle with equivalent aerodynamic properties).
Intermodal size range (PM 1-2.5 ) calculations in this study were made using chemical composition data obtained from the PM 1 , PM 1-2.5 , and PM 2.5-10 filter samples.The following assumptions were made for the calculations: 1) Intermodal aerosol is composed of coarse and fine mode aerosol with the two modes overlapping in the intermodal size range; 2) Both coarse and fine modes are chemically homogeneous over their respective particle size ranges; 3) Coarse mode aerosol is contained within the 1 to > 10 μm size range, and the chemical composition of the coarse mode aerosol is fairly represented by that of PM 2.5-10 ; and 4) Fine mode aerosol is contained within the 0 to 2.5 μm size range, and the chemical composition of fine mode aerosol in fairly represented by that of PM 1 (i.e., < 1 μm).
Based upon these assumptions, it is possible to calculate how much of the intermodal particle mass was contributed by the coarse mode PM (wind-blown dust) and how much by the fine mode PM (gas reaction or condensation prod-ucts).The following algorithms were developed: where X = mass contribution from coarse mode (μg/m 3 ), Y = mass contribution from fine mode (µg/m 3 ), I = PM 1-2.5 mass (μg/m 3 ), %S C = percent of species in PM 2.5-10 (used to represent the coarse mode), %S F = percent of species in PM 1 (used to represent the fine mode), %S I = percent of species in PM 1-2.5 (considered as the intermediate range).Equations ( 7) and ( 8) can easily be solved for X and Y, since X and Y are the only two unknowns.The percent of any specie " i " in the intermediate range that is contributed by the coarse, (%X) i , and fine, (%Y) i , modes can then be calculated.For the coarse mode contribution: And for the fine mode contribution: Using equations ( 9) and ( 10), %X = (7.7 − 35.6)/(2.1 − 35.6) = 83% contribution from coarse mode and %Y = (7.7 − 2.1)/(35.6 − 2.1) = 17% contribution from fine mode Thus, this example indicates that 83% of the SO 4 -2 particles in the intermediate region (sizes from 1.0 to 2.5 μm) are from the lower tail of the coarse mode and 17% are from the upper tail of the fine mode.
The data gathered from the trichotomous samplers from May to October, 1995, was analyzed using the method laid out in the above sample calculation.Two sets of data were chosen for analysis; 1) data from six days in July, 1995, was analyzed by ion chromatography (IC) and 2) data from five days in May, 1995, was analyzed by proton-induced X-ray emission (PIXE).
IC analyses were used to determine concentrations for the anions SO 4 −2 , NO 3 − , and Cl − .Table 3 presents the cal-culated percent contributions of coarse and fine mode PM to the intermediate size PM (PM 1-2.5 ), based upon SO 4 -2 concentration data for six July days of high, medium, and low mass concentrations.The results indicate that the coarse mode contribution to the intermodal size varies from 75% to 91% (with an average of 83%), and therefore, the fine mode contributions ranges from 9% to 25% (with an average of 17%).Due to their instability and associated analytical problems, the contribution analysis of NO 3 -and Cl -were found to be inconclusive, and are not shown.
The PIXE results indicated that Si, S, Ca, Fe, K, and Al were among the most abundant species in the samples analyzed.Therefore, they were chosen for the elemental contribution analysis in this study.The K and Al contribution results were questionable and thus excluded.Elemental concentrations for the five days in May are shown in Table 4. Table 5 shows the contribution results from analysis of the data in Table 4.
The results in

Data quality assessment
As with any field project, some sampling and analytical problems were encountered in this study.To ascertain that the data presented here is statistically valid, certain data reduction procedures were carefully followed in addition to the standard laboratory quality control measures.For example, no apparently-out-of-line data (i.e., does not fall within the 95% confidence limits) was invalidated, unless the cause was known or obvious (e.g., power failure).When two samplers were run, mass concentrations were averaged between both samplers as well as between two filters of the same particle size, so that the data are statis- tically more representative.
Table 6 shows precision data for the two co-located trichotomous samplers.The EPA performance specifications call for precision in the range of ± 5 μg/m 3 for PM 10 concentrations less than 80 μg/m 3 and ± 7% for concentrations over 80 μg/m 3 .

Conclusions
Results of the Phoenix PM 10 /PM 2.5 /PM 1 study using the trichotomous sampler technique can be summarized as fol-lows: First, it can be shown that most of the PM 10 in this area is made up of coarse mode aerosol (PM 2.5-10 ).Secondly, most of the intermediate size aerosol (PM 1-2.5 ) was due to the coarse mode (dispersion) aerosol.Therefore, crustal materials from sources such as windblown dust are the most significant contributor to the PM concentration in this area, far more so than the combustion related sources.Based up on the calculations presented, it appears that PM measurement using a cutpoint of 1 μm would capture about 90% as much fine mode mass as those made using a cutpoint of 2.5 μm, and would minimize any interference from natural sources (e.g., windblown dust).The reader should bear in mind that these results and conclusions are only valid for conditions similar to those at the Phoenix site, namely the dry, arid Southwest.

Table 2
Phoenix PM data expressed as percentages of PM 10

Table 5
clearly indicate that most of the PM in the intermediate region is associated with coarse mode particulate.

Table 3
Sulfate and mass data for six July days (μg/m 3 ) X* is the calculated coarse mode contribution to PM 1-2.5 (See text).

Table 5
Contribution to the intermodal (PM 1-2.5 ) aerosol from coarse and fine particle modes Average for the five days inMay, 1995