Characterization of Mainstream Cigarette Smoke Particle Size Distributions from Commercial Cigarettes Using a DMS500 Fast Particulate Spectrometer and Smoking Cycle Simulator

Abstract

Particle size distribution and number concentration measurements of mainstream cigarette smoke are reported for commercial cigarettes encompassing a broad range of design parameters. Measurements were made using a Cambustion DMS500 fast particulate spectrometer. Twenty-nine brand styles were evaluated using a 60-mL puff of 2-s duration taken once every 30 s. A subset of cigarettes was evaluated using additional smoking regimens to explore the influence of puff volume and filter ventilation blocking. The DMS500-derived particulate matter mass was compared with filter-collected mass to assess the reliability of the aerosol measurements. Under the 60-mL/2-s puffing condition, all puffs for all products were observed to exhibit count median diameters between 145 nm and 189 nm. Measured particle size was 12–22 nm smaller for a 60-mL puff relative to a 35-mL puff. Partial or complete filter ventilation blocking under the 60-mL/2-s puffing condition had a small effect on particle size. Some trends in particle size as a function of puff number and smoking regimen appear consistent with a tobacco-rod residence time/coagulation hypothesis; however, other observations suggest that smoke formation processes in addition to coagulation influence particle size. The DMS500 underestimates smoke particulate mass relative to gravimetric filter collection, indicating evaporation of cigarette smoke particulate matter within the instrument. Approximately 75% of the evaporated mass can be attributed to particulate phase water. Some data also suggest a possible underestimation of number concentration. This introduces a significant confounding bias in the measurements and limits the information on smoke formation that can be extracted.

[Supplementary materials are available for this article. Go to the publisher's online edition of Aerosol Science and Technology to view the free supplementary files.]

1. INTRODUCTION

The evaluation of measurement procedures and experimental results presented in this report is based upon certain aspects of the physicochemical nature of mainstream cigarette smoke (MSS) generally agreed upon by researchers in the field. A brief review of what is known about the physical properties of mainstream smoke and of certain specialized collection procedures will aid the interpretation of the results reported here and is presented in the following.

Of the large number of MSS chemical constituents, the majority are found in the particulate phase of the aerosol. The bulk of this particulate matter is formed by pyrolysis and distillation in a region located in the tobacco rod immediately downstream of the fire cone, or combustion zone, of the cigarette. Baker (1999) provides a comprehensive review of the MSS formation process, and a brief summary based on his description is provided here. Two distinct reaction regions are observed to exist in a tobacco rod during a puff: an upstream combustion zone at and on the periphery of the fire cone and a downstream pyrolysis/distillation region. In the exothermic combustion zone, oxygen drawn into the cigarette reacts with carbonized tobacco, producing primarily CO, CO₂, H₂O, and large amounts of heat. Heat is transferred by puff flow downstream to partially decomposed tobacco, where the bulk of the smoke particulate components are formed within the endothermic distillation/pyrolysis region. A concentrated vapor at around 600°C is produced, consisting of both components that have been formed by pyrolysis and decomposition of tobacco constituents and others that have transferred chemically intact from tobacco. The vapor cools rapidly as it flows downstream, and rapid saturation and particulate matter formation by condensation takes place. It has been suggested that tobacco undergoing pyrolysis may release solid, nonvolatile particles that serve as nuclei; however, saturation ratios are expected to be more than sufficient to produce homogeneous nucleation in the absence of nuclei. Significant downstream vapor deposition to unburned tobacco also takes place in the rapidly cooling flow. As the cooled smoke is transported through the tobacco rod, additional mass exchange between the volatile aerosol and surfaces can take place and coagulation within the concentrated aerosol can proceed. The aerosol exiting the filter is characterized as having particulate matter consisting of spherical liquid droplets of near unit-density and containing both dissolved and suspended substances dispersed in a complex gaseous mixture (Keith 1982; Davies 1988; Lipowicz 1988; Chen et al. 1990).

The mass of particulate matter in MSS is often equated with the mass determined by a gravimetric filter collection procedure referred to as the Cambridge pad method. The definition of smoke total particulate matter, TPM, and of smoke “tar” is based on this procedure. The method uses a high-efficiency glass fiber filter and corresponding filter holder through which smoke is drawn and captured. Mass is determined by weight difference before and after smoking. Material captured on the filter is described as TPM and is taken as a measure of the mass of the condensed phase of the smoke, while that passing through the filter is assigned to the gas or vapor phase. The Cambridge pad is typically analyzed for water and nicotine, and a “tar” value is calculated by subtracting the mass contribution of these components from the TPM. TPM and “tar” may be reported on a whole-cigarette or a puff-by-puff basis.

The particle size distribution (PSD) and number concentration of MSS have been of interest to researchers in the fields of tobacco science, aerosol science, and respiratory toxicology for several decades (Harris and Kay 1959; Phalen et al. 1976; Chang et al. 1985; Davies 1988; Martonen 1992; Baker and Dixon 2006). However, accurate and reliable size measurements have historically presented a challenge, largely due to the dynamic nature of the MSS aerosol. The dynamic qualities of the smoke are attributable to a particle number density of the order of 10⁹–10¹⁰ particles cm⁻³ and the presence of components across a continuum of volatilities from pure gases to nonvolatile particulate materials. The high number concentration results in significant, rapid changes in PSD of the undiluted smoke due to coagulation. Substantial dilution of fresh smoke is often required both to minimize the effects of coagulation and to provide aerosol concentrations within the operational range of some types of measurement instrumentation. The presence of volatile components in MSS can result in evaporation of significant amounts of particulate matter upon dilution (Hinds 1978; Chen et al. 1990), producing smoke PSD measurements not reflective of the PSD exiting the cigarette. Additionally, the submicron particle size of MSS and its relatively narrow distribution of particle sizes challenge the size range and resolution capabilities of some traditional measurement methodologies, e.g., cascade impactors (Hinds 1978). Single-particle light scattering measurements of MSS require substantial dilution levels and are subject to the evaporation issues discussed above. Ensemble scattering measurements on MSS have been accomplished by eliminating multiple scattering by making measurements on a narrow stream of smoke (Ingebrethsen 1986a), but require an assumed form for the size distribution in order to extract size parameters by the required data inversion on the measurements.

Ingebrethsen (1986b) and Bernstein (2004) have reviewed various MSS size measurement studies. A wide range of MSS size distributions have been reported, with average diameters on a count basis ranging from approximately 160 nm to 600 nm. The variety of techniques, dilution ratios, and aging times utilized in these studies are contributing factors to the lack of agreement among the measurements. In some cases, the use of aerosol sizing equipment not well suited for MSS measurement resulted in questionable particle size measurements. The more reliable measurements are in good general agreement, indicating that MSS count-averaged diameters fall within a range of 160–300 nm (Ishizu et al. 1978; Ingebrethsen 1986a, 1986b; Bernstein 2004).

More recently, Adam et al. (2009) used a differential mobility spectrometer (DMS500) manufactured by Cambustion Ltd. (Cambridge, UK) with a matter engineering smoking engine to measure the smoke PSDs produced by research cigarettes having a range of filter ventilation levels smoked under two puffing regimens. They reported count median diameters (CMD) ranging from 180-nm to 280-nm electric mobility diameter and particle number concentrations ranging from 1.4 × 10⁹ particles cm⁻³ to 1.5 × 10¹⁰ particles cm⁻³. Kane et al. (2010) used a custom-designed smoke sampling system and a Dekati Ltd. (Tampere, Finland) electrical low-pressure impactor (ELPI) to measure the PSD and number concentration of reference cigarettes. CMDs reported in their study were in the range of 140- to 240-nm aerodynamic diameter. Because MSS particulate matter is composed of spherical particles of approximate unit-density, both count-based electric mobility and aerodynamic diameters well describe the actual geometric diameter. Both studies reported a decrease in CMD and an increase in particle number concentration for conditions that reduced smoke residence time in the tobacco rod. These conditions include increased puff number, decreased filter ventilation, and increased average volumetric flow rates through the tobacco rod brought about by changes in puff volume. Both sets of authors attributed the reduced average sizes to diminished coagulation effects with decreasing smoke residence time in the tobacco rod.

The data in the present study were obtained by smoking a variety of commercially available cigarettes encompassing a wide range of “tar” yields, filter ventilation levels, and total cigarette lengths and circumferences. Aerosol measurements were made with a Cambustion DMS500, coupled with a Cambustion smoking cycle simulator (SCS). Twenty-nine brand styles were smoked using a single smoking regimen to explore the extent of variation in PSD and number concentration among the brand styles. A subset of seven products was studied using additional smoking regimens in order to explore the influence of puffing flow rate on particle size and number concentration. Also presented is a comparison of the DMS500-derived particulate mass measurements, on a whole-cigarette and a puff-by-puff basis, with gravimetric Cambridge filter pad measurements. This broad market survey employing multiple smoking conditions and direct comparison of the aerosol mass measurements to gravimetric values provides information not previously available. However, it is important to recognize that machine-smoked aerosol measurements do not provide the best input for dosimetry models unless the effects of mouth aging during puffing have been replicated or taken into account theoretically (Higenbottam et al. 1980; Tobin and Sackner 1982; Robinson and Yu 1999; Bernstein 2004; Ingebrethsen and Alderman, 2011).

2. EXPERIMENTAL

Symonds et al. (2007) have described the DMS500 and thus only a brief summary of its function is given here. The DMS500 classifies particles in the 5- to 1000-nm range according to their electrical mobility. Signals from 22 isolated electrometers within the classifying column are processed to generate a 38-channel real-time number concentration versus size spectrum at 10-Hz time resolution. The SCS is an attachment for the DMS500 that introduces MSS aerosol into the mobility analyzer. Although arbitrary puff profiles can be generated, a half sine wave puff profile was used exclusively during this study.

A labyrinth seal cigarette holder, pinch valve, and orifice meter make up the SCS sampling head. Prior to puffing and during the inter-puff intervals, all airflow needed to satisfy the DMS500 flow requirement of 30 L/min is provided by an external blower via the SCS control unit. Flow through the cigarette under study is provided by reducing airflow from the SCS control unit, resulting in “makeup” airflow being drawn from the sampling head. Because the DMS500 samples at an essentially constant flow rate, the volumetric flow rate through the cigarette at any point corresponds to the difference between the DMS flow and the flow provided by the HEPA-filtered dilution air supplied from the control unit. During a puff, flow from the cigarette is measured according to the differential pressure drop across a calibrated internal orifice in the sampling head. An automated control system and software monitor the desired smoking profile at 12-Hz resolution. The external air flow that is mixed with MSS is carefully metered and recorded on a temporal basis so that time-dependent concentration data may be extracted. A pinch valve in the sampling head is closed between puffs via a solenoid to ensure that no flow is drawn through the cigarette. Puff volume errors are typically less than 2%.

Owing to the high sensitivity of the electrometers used in the DMS500, substantial dilution of the smoke after sampling is required for the measurements. Preliminary experiments revealed that reported particle number concentrations, nominally adjusted to an absolute basis for varying dilution levels, were in fact dilution level dependent. Reported number concentrations reached a maximum when the extent of dilution provided by the DMS500 internal rotating disk diluter was maximized or was 200:1 at the rotary diluter stage for the DMS500 coupled to a SCS. Separately, dilution at varying levels takes place at the SCS sampling head due to the dynamic puff and external air flow rates. Total MSS dilution ratios ranged from 3445:1 to 28,825:1 and from 1912:1 to 16,733:1 for a 35-mL and a 60-mL puff, respectively.

The DMS500 electrometers were set to a low-gain mode for all measurements. All data reported here were averaged over a complete puff. Due to slight temporal broadening of the signals as the aerosol traverses the classifying column, all 2-s puff data were averaged 3 s beyond the time point corresponding to initial detection. Particle CMD, number concentration, geometric standard deviation (GSD), and aerosol mass values presented here were calculated using manufacturer-supplied data reduction software. The large number of experiments in the current study extended for many weeks; hence, prior to and after each day's data collection, at least one replicate of a filtered, American blend, moderately ventilated (30%) “control” cigarette was smoked to verify consistent operation of the DMS500 and SCS. A 60/30/2 smoking regimen was used for these monitoring runs, with 28 replicates collected over the course of this study. 60/30/2 designates a 60-mL puff of 2-s duration taken at 30-s intervals, and this convention will be used throughout this paper to describe smoking regimen parameters. The target puff volume drawn by the SCS was routinely verified with an external flow meter.

Twenty-nine brand styles were smoked using the 60/30/2 smoking regimen and varied appreciably in overall length, circumference, and percent filter ventilation. Percent filter ventilation, measured here at nominal flow rate of 17.5 cm³/s using a Sodim Instrumentation SMI PDV ventilation meter, is the fraction of total flow drawn at the mouth end of the cigarette that enters the cigarette through perforations in the filter. Increasing filter ventilation results in reduced flow through the tobacco rod, reduced amounts of tobacco burned during a puff, and increased dilution of mainstream smoke with ambient air. Detailed variation in cigarette design parameters will be presented within the “Results” section of this paper. The 60/30/2 regimen was chosen based on a desire to maintain flow rates through the cigarette that best approximate the “Massachusetts regimen” (a 45-mL puff of 2-s duration taken every 30 s with 50% of the filter ventilation holes obstructed) without the use of ventilation blocking. However, the relative differences in flow rates between the 60/30/2 regimen and the “Massachusetts regimen” vary as a function of filter ventilation. This is demonstrated by noting a +29% and a –19% difference in total flow rate for 0% and 55% filter ventilation, respectively. A smaller subset of seven brand styles was smoked using the 60/30/2 regimen with both 50% and 100% ventilation blocking as well as using the 35/30/2 regimen with no filter ventilation blocking.

TABLE 1 Puff-by-puff data for a 30% filter-ventilated, American blend cigarette smoked using a 60/30/2 regimen. Decreasing CMD, increasing number concentration, and decreasing aerosol mass/gravimetric mass ratios as puff number increases are representative of all products evaluated. ± one standard deviation of 28 replicates is provided

Puff	CMD (nm)	Concentration (per cm^−3 × 10^9)	GSD	DMS500 mass (mg)	Gravimetric mass (mg)	Aerosol mass/ gravimetric mass
1	171 ± 7	4.23 ± 0.38	1.40 ± 0.01	1.00 ± 0.14	1.48	0.67
2	168 ± 8	4.03 ± 0.34	1.44 ± 0.01	1.04 ± 0.17	2.07	0.50
3	169 ± 6	4.26 ± 0.42	1.45 ± 0.01	1.10 ± 0.18	2.19	0.50
4	167 ± 7	4.51 ± 0.36	1.46 ± 0.01	1.15 ± 0.17	2.31	0.50
5	165 ± 7	4.83 ± 0.34	1.45 ± 0.01	1.18 ± 0.17	2.92	0.41
6	166 ± 8	5.07 ± 0.38	1.46 ± 0.02	1.28 ± 0.22	2.95	0.43
7	165 ± 8	5.53 ± 0.36	1.45 ± 0.01	1.35 ± 0.23	3.68	0.37
8	161 ± 8	5.87 ± 0.40	1.46 ± 0.01	1.36 ± 0.23	3.97	0.34
9	159 ± 6	6.22 ± 0.51	1.46 ± 0.02	1.37 ± 0.19	4.19	0.33
10	159 ± 8	6.51 ± 0.60	1.46 ± 0.02	1.46 ± 0.27	4.49	0.32
11	154 ± 6	6.87 ± 0.60	1.47 ± 0.02	1.42 ± 0.24	4.56	0.31
12	152 ± 5	7.40 ± 0.58	1.47 ± 0.01	1.41 ± 0.18	4.90	0.29

The results of five replicate smokings of each product/smoking regimen condition were averaged for all data points. All cigarettes were lit using an electric lighter. Standard, regulatory smoking of cigarettes involves puffing until a designated tobacco rod length is reached. All cigarettes evaluated with the DMS500 were smoked to a butt length that well approximates (within 2 mm) the appropriate regulatory butt length. Many US states now require that cigarette paper be fire standard compliant (ASTM International 2002), which is accomplished using narrow bands of a permeability-reducing material that are applied to cigarette paper to limit diffusion of oxygen to the burning coal. If left unattended, the fire standard-compliant cigarettes will self-extinguish once the smoldering coal has encountered one of the permeability-reducing bands. All cigarettes used in the current study were fire standard compliant.

Lancaster Laboratories (Lancaster, PA) provided per cigarette “tar”, water, and TPM measurements collected using the 60/30/2 smoking regimen averaged for 20 replicate measurements, with all cigarettes smoked to the appropriate regulatory butt length. Puff-by-puff resolved TPM data were collected for the 30% ventilated “control” cigarette under the 35/30/2, 60/30/2, and 60/30/2 with 100% ventilation blocking regimens using a Borgwaldt (Hamburg, Germany) RM20 rotary smoking machine. For the puff-by-puff measurements, ten equivalent numbered puffs were collected on a 44-mm Cambridge pad and TPM was determined gravimetrically. Pads were analyzed by Lancaster Laboratories for water and nicotine content.

Because comparisons of the DMS500 particulate mass measurements with Cambridge filter particulate mass measurements indicated that significant particulate matter evaporation may be taking place in the DMS500, the extent to which Cambridge filter particulate mass determinations might be biased by vapor adsorption was assessed by MSS collection and quantification using an electrostatic precipitator. Compared with flow through a fibrous filter bed, an electrostatic precipitator tube provides a minimal adsorptive surface to the MSS as particulate matter is removed and is expected to eliminate or greatly reduce any possible adsorption of vapor-phase materials. Lancaster Laboratories collected TPM data for two cigarette types using a Borgwaldt linear smoking machine with an electrostatic precipitator as the primary collector, backed by a standard Cambridge filter pad through which the particle-stripped MSS vapor phase was passed. The back-up filter provided an additional measure of the extent of MSS vapor adsorption by the Cambridge filter pad.

TABLE 2 Select design parameters illustrating the extent of variation among the 29 cigarette types. Numerical values in parentheses indicate nonstandard (∼24–25 mm) circumference values. Average CMD and number concentration (±one standard deviation, n = 5) values were collected using the 60/30/2 smoking regimen

ID	Length (mm)	%Ventilation ^a	“Tar” (mg) ^b	CMD (nm)	Concentration (per cm^−3 × 10^9)
1	83	56	16.2	174 ± 2	3.32 ± 0.42
2	83	54	17.6	168 ± 3	4.44 ± 0.35
3 (23 mm)	99	58	18.3	173 ± 3	3.52 ± 0.24
4	83	42	19.0	166 ± 2	4.96 ± 0.42
5	97	53	19.9	174 ± 2	3.44 ± 0.48
6	83	29	23.1	160 ± 1	6.21 ± 0.30
7	83	18	23.2	166 ± 1	5.83 ± 0.79
8	79	31	25.2	159 ± 6	5.62 ± 0.33
9	83	33	25.3	168 ± 2	5.94 ± 0.46
10	97	26	25.6	165 ± 3	5.79 ± 0.27
11	83	19	26.8	164 ± 4	6.67 ± 0.46
12	83	30	27.0	161 ± 3	6.06 ± 0.43
13	97	43	27.8	161 ± 2	5.66 ± 0.32
14 (17 mm)	120	40	28.1	152 ± 4	6.68 ± 0.26
15	98	22	28.2	171 ± 3	5.20 ± 0.45
16	83	33	28.9	157 ± 3	6.28 ± 0.38
17	83	19	29.7	163 ± 2	6.96 ± 0.57
18	83	30	30.9	161 ± 4	6.29 ± 0.39
19	98	19	33.8	172 ± 2	6.70 ± 0.32
20	119	50	34.8	167 ± 3	5.95 ± 0.67
21	83	16	34.9	167 ± 3	7.56 ± 0.46
22	79	12	35.3	163 ± 2	8.54 ± 0.75
23 (21 mm)	120	44	36.3	167 ± 3	5.44 ± 0.53
24	83	21	37.6	160 ± 3	7.67 ± 0.32
25	80	0	38.9	164 ± 2	8.06 ± 0.48
26 (27 mm)	79	10	39.1	160 ± 1	7.37 ± 0.62
27	83	8	40.4	162 ± 3	7.41 ± 0.83
28 (27 mm)	78	0	40.4	162 ± 3	7.09 ± 0.27
29	69	Nonfilter	47.9	167 ± 4	10.6 ± 0.56

3. RESULTS

3.1. Puff-by-Puff Mainstream Smoke Particle Size and Concentration – General Trends

Table 1 lists CMD, particle number concentration, and GSD for 28 replicate measurements of the 30% filter ventilation American blend “control” cigarette on a puff-by-puff basis. The DMS500-derived and gravimetrically determined TPM mass is provided for each puff, along with the ratios of these two measurements. Standard deviations are provided where available. The CMD data show an approximately linear decrease (r ²  =  0.92) with increasing puff number. A decrease in CMD from 171 nm to 152 nm over the first 12 puffs represents a relatively small decrease relative to the confidence limits of the measurements. However, ANOVA treatment of the data reveals some statistically significant differences among puffs. When comparing all other puffs to puff 1, a significant decrease in particle size is observed for puff 4 and higher (p < 0.01). ANOVA analysis indicates a generally insignificant difference in CMD between all consecutive puffs. An indication that particulate matter may be evaporating within the DMS500 is the progressive decrease in aerosol mass/gravimetric mass ratio with increasing puff number, as presented in Table 1. Consequently, before attributing puff-by-puff trends in the DMS500-reported average size to a simple coagulation/residence time hypothesis, other changes in the properties of the smoke with puff number, and their interactions with the measurement procedure, will have to be evaluated.

A near-linear (r ²  =  0.99) increase in the DMS500 particle number concentration is observed from puff 2 to puff 12 following a slight decrease between puff 1 and puff 2. The increase in number concentration with puff number is supported by ANOVA analysis, which indicates a significant difference between the minimum value for puff 2 and that for puff 4 and greater. Additionally, ANOVA indicates that particle number concentrations are statistically different between consecutive puffs in most cases. The GSD values provided in Table 1 show a minimum value of 1.40 for puff 1 and values between 1.44 and 1.47 for the remaining puffs, without a clear trend with puff number. The puff-wise directional trends just discussed for the “control” cigarette are representative of those observed for all products studied, although detailed ANOVA analysis was not performed for all products.

3.2. Particle Size and Concentration Variability Under a Fixed Smoking Regimen

In order to demonstrate the wide variety of market products included in this study, Table 2 list total cigarette lengths, percent filter ventilation levels, and “tar” yields collected using the 60/30/2 smoking regimen for cigarettes included in the study. Data are also included for an unfiltered cigarette and for cigarettes not of approximate standard, i.e., 24- to 25-mm circumference. Average CMD and particle number concentration values reported by the DMS500 under the 60/30/2 smoking regimen are also provided in Table 2. The CMDs represent median diameters of all particles for all puffs calculated on a count-weighted basis. Average number concentrations were determined by dividing the cumulative particle count per cigarette by the total volume of smoke produced by each cigarette. Note that the average CMDs for all products lie between 152 nm and 174 nm. Thus, the average CMD values for all 29 cigarettes vary over approximately the same range as the puff-by-puff values for the single cigarette provided in Table 1. A table included as online Supplemental Information (Table S1) provides maximum and minimum individual puff CMDs and number concentrations measured for each cigarette. Average volume median diameters and gravimetric TPM for each cigarette measured using the 60/30/2 regimen are also provided. As observed from Table S1, average CMDs of all puffs for all 29 brand styles fall between 145 nm and 189 nm, i.e., the lowest per puff minimum and the highest per puff maximum value, respectively. Therefore, even though products evaluated in this data set encompass a broad range of “tar” yields, cigarette dimensions, and filter ventilation levels, the observed particle sizes for all products under this smoking regimen are relatively similar. No attempt was made to correlate particle size in this data set with the myriad of interactive smoke formation and transport variables that result from such a broad array of smoke construction parameters.

The average number concentrations for the filtered products listed in Table 2 range between 3.32 × 10⁹ particles cm⁻³ and 8.54 × 10⁹ particles cm⁻³, and the nonfiltered product shows an average number concentration of 10.6 × 10⁹ particles cm⁻³. Excluding the nonfiltered product, per puff minimum and maximum number concentrations compiled in Table S1 range between 1.7 × 10⁹ particles cm⁻³ and 10.2 × 10⁹ particles cm⁻³.

Adam et al. (2009) reported that MSS particle size is well represented by a log-normal distribution and this was observed to be the case for all measurements in the current study. No clear trends in GSD as a function of puff number were apparent within data, with the exception that the first puff GSD values were generally in the lower end of the observed range (1.39–1.42). Due to the consistency among all measured GSD values, a detailed individual cigarette listing has not been provided. The vast majority of distributions (post-puff 1) measured during this study had GSDs in the range 1.45–1.47. Essentially, all particles making up each MSS PSD encountered in this study fell between 40-nm and 600-nm electric mobility diameter.

3.3. Influence of Puff Volume and Ventilation Blocking

Figure 1a illustrates the effect of smoking regimen on particle size (number-weighted CMD of all five replicates) for a subset of seven products. The four smoking regimens employed were 35/30/2 and 60/30/2 with no ventilation hole blocking, and 60/30/2 with both 50% and 100% ventilation blocking. These cigarettes are labeled according to the ID number provided in Table 2. With the exception of product ID 14, all total cigarette lengths were between 79 mm and 83 mm and were of approximate standard circumference. Product ID 14, a super-slim-type cigarette of 120-mm total length, was included here since its 60/30/2 smoking regimen resulted in the smallest observed CMDs. Product ID 1was included because it yielded CMDs among the largest measured. The remaining products listed in Figure 1a provide a representative sampling of ventilation levels.

It is evident from Figure 1a that the DMS500-reported average sizes are smaller for each product under the higher-volume 60/30/2 regimen as compared with the lower-volume 35/30/2 regimen. Decreases in CMD of 12–22 nm were deemed significant via ANOVA analysis at the 95% confidence interval, consistent with the hypothesis that decreasing tobacco-rod residence time brought about by the higher-volume puff results in smaller particle sizes. However, other data in Figure 1a yield examples of how factors other than residence time play a role in determining particle size. Residence time considerations alone would predict significant decreases in particle size between the 60/30/2 smoking regimen with no ventilation blocking and the 60/30/2 regimen with 50% ventilation blocking, as decreased filter ventilation also serves to increase the flow rate of MSS through the tobacco column. Further decreases in CMD would be expected with the 60/30/2 regimen plus 100% ventilation blocking. But as illustrated in Figure 1a, only two of the seven products evaluated (ID 1 and 14) clearly display the decrease in average size that would be expected if residence time were the dominant determinant of average particle size.

Product IDs 1 and 14, with 56% and 40% filter ventilation, respectively, seem to result in a reduction in size as the residence time hypothesis would predict. These relatively highly ventilated products would be expected to have their flow rates most affected by ventilation blocking. However, data for product ID 4, with 54% ventilation, does not show the expect trend. The degree of filter ventilation for the remaining products spans 12–30%. A separate factor that also changes with ventilation blocking, and one that must be considered in any attempt to provide a mechanistic explanation for observed trends in average size, is the fact that the fraction of water measured in the MSS TPM increases with increasing puff volume and extent of ventilation blocking. This can potentially affect the degree of particle evaporation during the DMS500 measurements. This latter point will be discussed further in the following.

FIG. 1 Plots depicting variation in (a) CMD and (b) particle number concentration among several smoking regimens. 50% VB and 100% VB indicate the extent of filter ventilation hole blocking. The ID numbers found below each grouping correspond to cigarettes identified in Table 2.

The effect of smoking regimen on average particle number concentration is illustrated in Figure 1b, where it is clear that significant increases in number concentrations generally result under the more intense regimens. Particle concentration was higher by a factor of 10–60% for the 60/30/2 regimen compared with the 35/30/2 regimen for the seven-product subset. Similar increases were noted when comparing the 60/30/2 regimen with the 60/30/2 regimen plus 50% ventilation blocking. Larger increases, 12–136%, were measured when comparing the 60/30/2 regimen with the 60/30/2 regimen plus 100% ventilation blocking. The increase in number concentration with increased flow down the tobacco rod, achieved by both increased puff volume and ventilation blocking, is much more consistent and significant than the changes in average size presented in Figure 1b, with only product IDs 21 and 22 exhibiting a slight deviation from the trend, going from 50% to 100% ventilation blocking.

3.4. Comparison of DMS500 Mass Measurements with Gravimetric Measurements

A reconciliation of aerosol particulate mass measured by gravimetric filter collection with that calculated from aerosol size and number measurements is often problematic due to the difficulty in simultaneously obtaining accurate absolute size and number concentrations and the frequent occurrence of artifacts introduced in filter collection. This task is even more challenging when dealing with an aerosol such as MSS, which contains substantial amounts of volatile smoke components. Nevertheless, because of the central role of TPM measurements in characterizing MSS, and with the goal of making a rigorous assessment of the current results, a detailed comparison of gravimetric and DMS500-derived mass determinations is warranted. In the vast majority of cases, TPM mass determined from the DMS500 measurements was less than that indicated by gravimetric TPM mass measurement.

A reasonable possibility for the discrepancy between gravimetric and DMS500-derived TPM mass ratios is that some material measured as particulate mass by gravimetric filter collection evaporates as a result of the relatively high dilution level required for SCS operation and DMS500 measurement, the reduced pressure of ∼25 kPa in the DMS500 column, and a temperature of 55°C maintained in the DMS500 cyclone. Particle evaporation under these conditions would result in an underestimation of both the average size indicated by the DMS500 and the calculated particle mass relative to the gravimetric measurement on undiluted smoke.

The degree to which the lower particulate mass measured by the DMS500, relative to the Cambridge filter method, indicates particle evaporation in the instrument depends upon the accuracy with which the filter collection method measures the particulate mass in the undiluted aerosol. Both vapor adsorption and deposit evaporation during collection are possibilities with aerosols having volatile particulate matter. Electrostatic precipitation measurements were conducted on an American blend cigarette with 30% filter ventilation and on an Eclipse cigarette (Foy et al. 2009) that primarily heats, rather than burns tobacco, both smoked under the 60/30/2 smoking regimen. Twenty replicate cigarettes were smoked for each collection. For the tobacco-burning cigarette, 36.5 mg of TPM per cigarette were captured in the electrostatic precipitation tube compared with 39.1 mg of TPM per cigarette for identical smoke with Cambridge filter collection. For the Eclipse cigarette, which contains a substantially larger fraction of water in the TPM than tobacco-burning cigarettes, 29.0 mg of TPM were captured in the electrostatic precipitator tube compared with ∼50 mg of TPM per cigarette routinely quantified by Cambridge filter collection. No measurable mass was collected on the precipitator back-up Cambridge filter for either cigarette. The absence of quantifiable mass on the back-up filter and the agreement between the two gravimetric methods indicated a low probability of significant vapor adsorption or deposition evaporation artifacts in the particulate mass determinations by Cambridge filter collection for the tobacco-burning cigarettes of this study. The discrepancy between Cambridge pad-collected and electrostatic precipitator-collected Eclipse MSS suggests the possibility of significant water vapor adsorption during the direct collection of the Eclipse smoke on a Cambridge pad. This observation may be related to the very high glycerol content of the Eclipse smoke compared with tobacco-burning MSS and the much larger puff-by-puff variation in TPM yields for the Eclipse cigarettes. Loading of the Cambridge pad with glycerol may create a highly hydroscopic surface that can collect water vapors, from the cigarette and from humidity-controlled laboratory air drawn through the cigarette, on subsequent puffs that have much lower TPM mass levels than earlier puffs. Additionally, special lighting procedures associated with the Eclipse smoking may also contribute to the difference in electrostatic and gravimetric TPM measurement. The fact that the backup pad on the electrostatic precipitator collector does not become loaded with glycerol and does not collect measurable water vapors would be consistent with such a hypothesis. Furthermore, it should be pointed out that the 29.0-mg TPM measured for the Eclipse cigarette by the electrostatic precipitator method exceeds the particle mass quantified by the DMS500 by a factor of ∼10.

The likelihood of significant evaporation of volatile particulate matter components during the DSM500 measurements is supported by the data in Figure 2. Figure 2 plots the ratio of the DMS500-derived particulate mass to gravimetrically determined particulate mass versus the mass fraction of water in TPM. All data in Figure 2 represent complete smoking of the cigarette replicates and were obtained using the 60/30/2 smoking regimen without ventilation blocking. Twenty-two of the 29 brands evaluated are included in the plot, along with an Eclipse cigarette. Some brand styles were not included due to an unacceptable difference between the number of puffs supplied to the DMS500 and the number used to obtain the gravimetric measurement. Criteria for inclusion in the plot were agreements within ±0.7 puffs. For tobacco-burning cigarettes, the DSM500-derived aerosol mass/gravimetric mass ratios range from 0.24 to 0.61. For the Eclipse cigarette, which was smoked to 22 puffs, the ratio is 0.06. It is also evident from Figure 2 that the DMS500 aerosol mass to gravimetric mass ratio decreases in a linear fashion as the fraction of water in the TPM increases. Figure 2 is consistent with the hypothesis that the DMS500 underestimates MSS particle mass due to particle evaporation at high dilutions and low pressures within the instrument. Loss of water alone for the tobacco-burning cigarettes of Figure 2 would result in a 12–35% underestimation of particulate mass by the DMS500 under the 60/30/2 regimen. Since the underestimation of TPM by the DMS500 is greater than that attributable to water evaporation alone, but still strongly correlated with water content of the TPM, the water content is likely an indicator of volatile particulate mass level.

FIG. 2 The ratio of particulate matter mass measured by the DMS500 to Cambridge pad TPM measurements plotted as a function of the water fraction of TPM. These data were collected using a 60/30/2 smoking regimen and correspond to complete smoking of each cigarette.

If it is hypothesized that the mass fraction of fresh TPM that evaporates in the DMS500, F_v , is linearly related to the mass fraction of water in the TPM, F_w , by a ratio r, then F_v =rF_w . Using this hypothesis and equating gravimetric mass to fresh TPM, it is predicted that aerosol mass divided by gravimetric mass yields 1−rF_w and that the plot in Figure 2 should be a straight line with slope –r. The linearity in Figure 2 indicates good agreement with this simple model over a broad range of cigarette types and yields r = 1.25, with r calculated excluding the Eclipse data point in order to compare only tobacco-burning cigarettes. This r value suggests that the fraction of fresh MSS TPM that evaporates in the DMS500 is predominantly water and varies with the cigarette type. The data suggest that additional evaporation of approximately 25% more than that predicted value based on water-alone evaporation takes place, indicating a 15–44% underestimation of particulate mass by the DMS500 for tobacco-burning cigarettes attributable to evaporation. A plot of aerosol mass divided by dry TPM, or TPM minus water, is not predicted to yield a simple linear relationship under the hypothesis stated above.

There is also support in the data for a possible smoke concentration effect on measured aerosol mass/gravimetric mass ratios. As illustrated in Figure 3, the water content of TPM is observed to decrease as filter ventilation increases. Although not depicted, a plot of filter ventilation versus average particle concentration taken from Table 2 reveals that number concentration is inversely proportional to filter ventilation. Thus, MSSs with the highest particulate matter number concentrations, i.e., lowest filter ventilation levels, also have higher TPM water content and lower aerosol mass/gravimetric mass ratios. Taken together, there is a suggestion that aerosol mass/gravimetric mass ratios are inversely dependent on particle number concentration. The effect of number concentration would be through the hypothesis that particle number concentration measurements are underreported to a greater extent with increasing actual particle number concentrations.

FIG. 3 Correlation between the particulate matter water fraction and percent filter ventilation.

The effect of number concentration can be assessed to some extent by adjusting the dilution level of the instrument's internal rotating disk diluter. If dilution levels are accurately applied and accounted for in the data analysis, reported particle number concentrations will ideally be equal for identical aerosol streams measured at different rotating disk dilution levels. However, the DMS500 consistently indicates higher particle number concentrations for theoretically identical aerosols at higher internal dilution values. Table 3 provides data illustrating the magnitude of this effect on aerosol mass/gravimetric mass ratios. The comparisons are for the fourth and eighth puffs of six averaged replicates of 54% and 21% filter-ventilated commercial cigarettes smoked under the 35/60/2 regimen. The filter exit character of the smoke particulate matter, i.e., the TPM water content and particle number concentration of the smoke exiting the cigarette, is expected to be essentially equal for each puff number–ventilation level pair listed in Table 3, thereby isolating the effect of secondary dilution level. These data reveal 8–14% higher aerosol mass/gravimetric mass ratios for the 200X versus 75X dilution level. This can be attributed to an increased particle counting rate at the higher secondary dilution levels, as measured CMD and GSD did not significantly vary with secondary dilution level. The observation that CMD is independent of the secondary dilution level provides some support to the critical dilution hypothesis offered by Chen et al. (1990), i.e., there exists a level of dilution above which no further rapid decrease in smoke particle size due to evaporative loss of TPM components occurs. Chen et al. (1990) suggest that the critical dilution ratio is likely between 10 and 100.

TABLE 3 Aerosol mass/gravimetric mass ratios for various puff number/cigarette type pairs at DMS500 secondary dilution ratios of 75X–200X. Data were collected using the 35/60/2 smoking regimen

Secondary dilution	Puff 4 21% vent.	Puff 8 21% vent.	Puff 4 54% vent.	Puff 8 54% vent.
75X	0.57	0.52	0.71	0.74
100X	0.60	0.54	0.74	0.73
150X	0.60	0.55	0.80	0.81
200X	0.70	0.60	0.80	0.83

Finally, Figure 4 presents a plot of the DMS500-derived particulate mass/gravimetric mass ratio versus puff number for a single cigarette type evaluated using three smoking regimens. The cigarette chosen was the “control” cigarette, for which some data are provided in Table 1 and Table 2 (ID 12). In general, both the particle number concentration and particulate matter water content increase with increasing puff numbers and more intense smoking regimens. A figure included as online Supplemental Information (Figure S1) is provided, depicting puff-by-puff collected TPM water fractions corresponding to the cigarette and smoking conditions indicated in Figure 4. Thus, the data presented in Figure 4 further demonstrate that the DMS500-derived mass/gravimetric mass ratios generally decrease with increasing particle number concentration and particulate matter water content.

FIG. 4 The influence of puff number and smoking regimen on aerosol mass/gravimetric mass ratios of a 30% filter-ventilated American blend cigarette. 100% VB indicates 100% filter ventilation blocking.

4. DISCUSSION

The findings of the current study address three areas of importance to researchers studying the aerosol properties of MSS and related topics. Firstly, an assessment of the accuracy and sources of error involved in the measurement of the PSD of MSS with the DMS500/SCS is provided based on measurements for a variety of cigarette types, under various conditions and with concurrent determinations of other MSS properties. Secondly, the results provide a broad survey of the PSD and number concentration of MSS from a variety of market cigarettes smoked under a range of regimens and evaluated within the uncertainties imposed by the DMS500 technique. Finally, through smoking regimen manipulation, puff-by-puff determinations, and other measurements, some limited evaluations of smoke formation and transport processes occurring in the cigarettes are derived from the present findings. These three general areas are interdependent, but will be discussed in turn in the following.

Several observations reported here support the not unexpected probability that the high dilution levels and reduced pressures produced in the DMS500/SCS result in a significant and variable degree of evaporation of MSS particulate mass during measurement. Although the support for this hypothesis depends most directly on the accuracy of the Cambridge filter method for fresh smoke, several other potential sources of systematic error should be mentioned before a full discussion of the gravimetric procedures is given.

The particle size measurement range sensitivity of the DMS500 appears to comfortably encompass all sizes present in MSS as judged by the specifications of the instrument and the literature reports of MSS particle size measured by a variety of techniques. Certainly, negligible counts were obtained for the smallest and largest size bins for all measurements. The DMS500 calculation of particle size requires an assumption of a particle density in order to convert the electrical mobility measurements to particle mass. A density value of 1.0 g/cm³ is assumed for this work. The range of particle density measurements reported in the literature (Lipowicz 1988; Chen et al. 1990) indicate that any error associated with our use of the 1.0 g/cm³ value cannot account for the large observed differences between the gravimetrically and aerosol-property-derived particulate matter mass data. Electric mobility measurements also assume that a well-characterized charge distribution has been achieved under the conditions of the measurement, i.e., the actual charge distribution in the measurement is accurately described by theory (Biskos et al. 2005a, 2005b). The validity of this latter assumption is difficult to address in the present study. Other potential sources of instrumental error that might influence PSD measurements and mass calculations include inaccurate dilution ratio and sampling rate flow measurements, as well as losses of particle mass during transit in the instrument.

The DMS500 measures both PSD and number concentration on an absolute basis and, in theory, allows a calculation of total particle mass that can be compared with independent gravimetric particulate mass measurements on replicate samples of smoke. Such comparisons in the present study indicate that total particulate mass calculated from the aerosol properties is significantly and to a variable degree lower than that measured gravimetrically on the fresh smoke. The hypothesis that the apparent underestimation of total mass by the DMS500 measurements is real and attributable to particle evaporation during the measurement is dependent on the degree to which the mass collected by the Cambridge filter method is an accurate measure of the total particulate mass present in fresh, undiluted MSS.

Since only puffed, whole, undiluted MSS is passed through the filter during Cambridge pad collection, there is no opportunity for evaporation of previously deposited particulate mass as is the case in continuous flow filter collection procedures. Additionally, after a small number of puffs have been collected, the filter surface is essentially coated with MSS TPM and the vapor–surface equilibrium is essentially identical to the vapor–particle equilibrium and there is no driving force for selective uptake or depletion of particulate components. Consequently, the mass collected by the Cambridge pad method is expected to be representative of material present in the particulate phase of the fresh, undiluted MSS and this expectation has been confirmed experimentally.

Collection of MSS TPM via electrostatic precipitation with a downstream Cambridge filter pad in place suggests that gravimetrically determined TPM is negligibly influenced by vapor condensation onto the pad. Higuchi et al. (2000) reports that <1% of TPM mass was collected on a back-up filter downstream of an electrostatic precipitator tube when an Eclipse cigarette was smoked under the 75/35/2 regimen. Torrence et al. (2002) reported slightly more TPM capture via electrostatic precipitation relative to a filter measurement for a Kentucky 1R4F reference cigarette smoked using the 35/60/2 regimen. Finally, experiments reported above for the current study confirm not only that negligible MSS vapor adsorption takes place on Cambridge pad filters, but also that particulate mass levels collected by electrostatic precipitation from tobacco-burning cigarette MSS is in excellent quantitative agreement with those determined by Cambridge filter collection.

With the reasonable assumption that the Cambridge filter gravimetric determination of particulate mass is an accurate measure of the total particulate mass present in undiluted MSS, the data of Figure 2 are a clear demonstration that the DMS500 reports size and number concentration measurements that only fractionally account for the total particulate mass present in the MSS aerosol sampled. Additionally, the fact that the fraction of fresh particulate mass accounted for by the DMS500 measurements drops in direct proportion to the fraction of volatile water in the MSS particulate matter strongly suggests that particulate evaporation with concomitant undersizing of the fresh particles is taking place in the instrument. Other data of the current study are also consistent indicators that significant MSS particle evaporation occurs in the DMS500. A number of factors, both construction parameters and smoking regimen characteristics, are reasonably expected to alter the chemical nature of the MSS particulate phase and produce variable changes in levels of evaporation that would be reported by the DMS500 as changes in particle size. The data presented here suggest that these differences, most specifically TPM water content, impact aerosol mass/gravimetric mass ratios and indicate a probable undersizing of fresh, undiluted MSS particle size by the DMS500 that varies with product design, smoking conditions, and puff number. Further complicating the interpretation of the data sets is the observation that the number concentration reported by the DMS500 varies with smoke concentration level, as assessed by comparing would-be identical cigarette puffs at multiple levels of total dilution.

The question then must be addressed as to what extent we can establish limits on actual MSS particle size for the cigarettes studied from the DMS500-reported average sizes and probable instrumental biases. It should first be emphasized that the very wide range of cigarette types and smoking regimens employed yielded a relatively narrow range of measured average particles sizes for all puffs of all cigarettes. The range of average sizes obtained from a reconciliation of aerosol mass with gravimetric mass by either size or number concentration adjustments described below still predicts a very narrow overall range of average particle size for the entire data set.

By assuming that the two principal causes of discrepancy between the DMS500 mass measurement and gravimetric measurement are size underestimation and/or number concentration underestimation, we can establish a reasonable range for the actual size and particle number concentration that yield the measurements reported by the DMS500. A select puff-by-puff size and number concentration measurement subset has been used to illustrate the probable range of uncertainties in the actual MSS particle sizes and number concentrations.

The data plotted in Figure 5a illustrate puff-by-puff resolved CMD values for the same cigarette (ID 12, Table 2) smoked under two regimens and determined in three ways: (1) values directly reported by the DMS500, (2) values calculated assuming that all particulate-phase water completely evaporated within the DMS500 prior to sizing and that water was the only component that evaporated, and (3) that particulate matter evaporated to the full extent required to reconcile aerosol mass and gravimetric mass, i.e., particle undersizing due to material evaporation within the DMS500, is the only source of measurement error and is labeled “water plus other” in the figure legend. Assuming complete evaporation of only water, the DMS500 may underestimate particle size by 26–42 nm and 13–51 nm under the 35/30/2 and 60/30/2 regimen, respectively, depending on puff number. This comparison excludes the first puff for the 35/30/2 regimen, where the DMS500 mass value was slightly greater than the gravimetric value. If it is assumed that particle undersizing by the DMS500 due to evaporation is the only source of error leading to fractional accountability of particulate mass, size adjustments upward of 32–55 nm and 24–78 nm are required for the 35/30/2 and 60/30/2 regimen, respectively, in order to reconcile the aerosol mass measurement with the gravimetric mass measurement.

FIG. 5 (a) CMD values reported directly by the DMS500, calculated assuming the evaporation of TPM water and only water, and calculated assuming that TPM evaporation occurred to the extent to fully account for aerosol mass/gravimetric mass ratios in Figure 4. (b) Number concentration data directly reported by the DMS500, calculated after adding TPM water back as particle mass and calculated assuming that number concentration underestimation occurred to the extent to fully account for aerosol mass/gravimetric mass ratios in Figure 4, i.e., no evaporation.

Owing to the cubic dependence of particle volume on particle diameter, a proportionally smaller adjustment in particle diameter, relative to the fractional mass accountability presented in Figure 4, is required to reconcile aerosol-property-calculated masses to gravimetric masses. To illustrate, a puff with a DMS500 mass/gravimetric mass ratio of 0.5 would correspond to a 26% underestimation of particle diameter relative to that based on the gravimetric mass values being the true suspended mass measure. Equation (1) better illustrates this relationship:

where _ADJ is the adjusted value diameter of average mass, _DMS is the DMS500-reported diameter of average mass (calculated via a Hatche–Choate equation, assuming a log-normal PSD), M _DMS and M _GRAV are the puff masses determined by the DMS500 and gravimetric method, respectively. The water-only particle size adjustments discussed above were made by adding the mass of water measured on a puff-by-puff basis to the M _DMS term in Equation (1).

One notable observation taken from Figure 5a is that particle size values remain essentially constant or increase slightly with increasing puff number when adjusted to agree with the gravimetric mass measurements. This is in clear contrast to trends observed in the current study for unadjusted data, as well as to puff-by-puff trends reported by most other published studies (Adam et al. 2009; Kane et al. 2010). Note that Adam et al. (2009) and Kane et al. (2010) used a 50X and a variable 500–1000X dilution factor, respectively, and that these values are considerably lower than the dilution values employed here. Also notable from Figure 5a is the overlap in CMDs among the latter puffs for both regimens.

A method for evaluating the possible extent of an underestimation of number concentration is discussed below and illustrated in Figure 5b. Here, three estimates of number concentrations are provided on a puff-by-puff basis: (1) as directly reported by the DMS500, (2) the number concentration required to reconcile the remaining difference between aerosol mass and gravimetric mass after experimentally determined water has been added back as aerosol mass, and (3) number concentration calculated assuming that the DMS500 CMDs reported are accurate, labeled in the figure legend as “no evaporation.” The magnitude of the upward adjustment in number concentration is directly proportional to the degree of error between the DMS500-derived puff mass and the gravimetric puff mass, e.g., a DMS500 mass accounting for 50% indicates an upward adjustment in number concentration of 100% relative to that reported directly by the DMS500. These estimates suggest that actual number concentrations may be higher than measured by the DMS500 by 45–93% and 25–133% for the 35/30/2 and 60/30/2 regimen, respectively, for the water-only evaporation calculation. The ranges of actual number concentrations are predicted to be 58–128% and 48–247% higher for the 35/30/2 and 60/30/2 regimen, respectively, by the calculations that assume that the DMS500 CMDs are accurate. The number concentration estimates required to reconcile the mass measurements are very high, particularly when the DMS500-reported particle sizes are taken to be explicitly accurate, and likely physically unrealistic on the time scales of the measurements due to coagulation. This suggests that the size adjustments shown in Figure 5a are probably a reasonable depiction of the range of fresh MSS average sizes indicated by the DMS500 measurements in light of the possible errors involved.

Finally, it is observed that only limited evaluations can be made of the current results based on mechanistic smoke formation and transport effects. Overall, the ranges of average particle size and number concentration are narrow for the entire set of cigarette types and smoking regimens studied. Also, uncertainties in the DMS500 measurements are of the order of variations in size observed across cigarette types and with smoking regimen changes. To complicate matters, many of the variables studied, e.g., construction parameters, puff numbers, puff volumes, etc., have multiple, interactive effects on the smoke produced and some of these effects appear to interact with the accuracy of the DMS500 measurements, most notably smoke particulate volatility. As an example of the complications encountered in providing mechanistic analyses of these types of data, the following can be considered. A 60/30/2 smoking regimen compared with a 35/30/2 regimen results in more tobacco consumed per puff, which reasonably leads to an increase in particle number concentration and material available for particle growth via condensation immediately downstream of the pyrolysis/distillation zone. Successive puffs in the smoking cycle not only act on the tobacco originally present within a given length of tobacco rod, but also on material generated during previous puffs that had condensed on downstream tobacco strands. Additionally, it is well known that changes in filter ventilation, smoking regimen (Counts et al. 2005), and puff number lead to differences in MSS TPM chemical composition. The data presented here suggest that these differences, most specifically TPM water content, impact aerosol mass/gravimetric mass ratios and indicate a probable undersizing of filter exit particle size by the DMS500 that varies with product design, smoking conditions, and puff number. Further complicating the interpretation of the data sets is the observation that the number concentration reported by the DMS500 varies with actual smoke concentration level, as assessed by comparing would-be identical cigarette puffs at multiple levels of total dilution.

5. CONCLUSIONS

Overall, the findings reported here regarding the range of particle sizes of MSS from a wide variety of cigarette types smoked under a variety of regimens are believed to be of unique value to researchers in related fields. When smoked under a single 60/30/2 smoking regimen, commercial cigarettes encompassing a broad range of “tar” yields, cigarette lengths, and filter ventilation levels produce fresh mainstream smoke with very similar PSDs. CMD values averaged over all puffs range from 152 nm to 174 nm, with the CMD for all puffs for all products falling between 145 nm and 189 nm. The results of smoking a smaller subset of cigarettes under regimens of varying intensity indicate a statistically significant decrease in particle size under the 60/30/2 smoking regimen relative to the 35/30/2 regimen (a 12- to 22-nm decrease in CMD under the more intense smoking regimen). This shift is qualitatively consistent with the expected flow rate, residence time, and coagulation effects, but may be affected by changes in particulate matter volatility and evaporation in the electrical mobility analyzer. A much smaller or insignificant decrease in particle size is observed when comparing the 60/30/2 regimen plus no ventilation blocking with the 60/30/2 regimen plus 50% or 100% ventilation blocking. As ventilation blocking of this magnitude serves to significantly increase residence time in the tobacco rod, other factors apparently play a role in final particle size. In essentially all instances, the DMS500 underestimated particulate matter mass as determined by filter collection. Data presented here support the hypotheses that some degree of evaporation of semi-volatile MSS components is occurring within the DMS500 and that the particle number concentration may be underestimated under some conditions by the DMS500. The water content of the particulate matter (implicated here as an evaporating species) as well as the DMS500-reported particle number concentration (which appear to be a function of the actual concentration) both co-vary with the construction parameters and smoking conditions and influence the DMS500 mass accountability, preventing a quantitative determination of the relative contributions of the two effects to the errors involved in the DMS500 PSDs and number concentrations reported for the measurement of fresh MSS. Limited new information on MSS formation and transport can be extracted from the current results due to the numerous complex interactions among the cigarette function and measurement procedure.

Notes

^aMeasured at 17.5 mL/s.

^bCollected using a 60/30/2 smoking regimen, n = 20.