Delivery Levels and Behavior of 1,3-Butadiene, Acrylonitrile, Benzene

For each of the 26 VOCs, the smoke PM from the two brands yielded very similar Kp values at 20 °C. In addition, the vapor pressure-dependent Kp value...
0 downloads 0 Views 132KB Size
Chem. Res. Toxicol. 2004, 17, 805-813

805

Delivery Levels and Behavior of 1,3-Butadiene, Acrylonitrile, Benzene, and Other Toxic Volatile Organic Compounds in Mainstream Tobacco Smoke from Two Brands of Commercial Cigarettes James F. Pankow,*,†,‡ Wentai Luo,‡ Ameer D. Tavakoli,‡ Cai Chen,‡ and Lorne M. Isabelle‡ OHSU Cancer Institute, and OGI School of Science & Engineering, Oregon Health & Science University, Portland, Oregon 97291 Received November 10, 2003

Mainstream tobacco smoke (MTS) was collected from Camel and Marlboro cigarettes for the determination of the delivery levels and equilibrium gas/particle partitioning constants Kp (m3 µg-1) of 26 volatile organic compounds (VOCs) of toxicological interest. Kp values are important for understanding the fractional distribution of each compound of interest between the gas and the particle phases of MTS. The experimental method involved (i) drawing a smoke sample into a Teflon sampling bag at 20 °C, (ii) allowing the smoke particulate matter (PM) to collect on the walls of the bag, (iii) sampling the bag to determine the initial gas phase concentration of each VOC, (iv) removing the gas phase from the bag, (v) refilling the bag with humidified nitrogen gas, (vi) reestablishing the gas/PM equilibrium, and (vii) redetermining the gas phase concentrations. For each smoke sample, a comparison of the initial and redetermined gas phase concentrations allowed calculation of the total (i.e., gas + particle) delivery level () mtot, ng cig-1) and Kp value () cp/cg) at 20 °C for each compound, where cp (ng µg-1) ) concentration in the PM phase and cg (ng m-3) ) concentration in the gas phase. Significant deliveries were observed for a number of carcinogenic VOCs. For the Camel cigarettes tested, the average mtot values for 1,3-butadiene, acrylonitrile, and benzene were 104.6, 104.4, and 104.8 ng cig-1, respectively; for Marlboro, the mtot values were 105.0, 104.6, and 104.7 ng cig-1, respectively. For each of the 26 VOCs, the smoke PM from the two brands yielded very similar Kp values at 20 °C. In addition, the vapor pressure-dependent Kp values of the 26 VOCs were in close agreement with predictions made by the Pankow theory of absorptive gas/ particle partitioning [Pankow, J. F. (1994) Atmos. Environ. 28, 185-188]. These results can be used in general predictions of chemical behavior in tobacco smoke, including deposition mechanisms and rates in the respiratory tract from inhaled MTS. Example calculations are provided to illustrate how the gas phase fraction at equilibrium (fg,e) increases strongly with increasing compound vapor pressure and temperature and with dilution of the inhaled tobacco smoke total PM concentration (µg m-3).

Introduction The mainstream tobacco smoke (MTS) that is produced by a burning cigarette and delivered by inhalation to a smoker is an aerosol system that is composed of a great many suspended smoke particles and a surrounding gas phase. All of the organic and inorganic compounds that are present in tobacco smoke tend to partition between the smoke particulate matter (PM) phase and the surrounding gas phase. Because this partitioning is compound-dependent, so too are the mechanisms, speeds, and locations of the respiratory tract (RT) deposition of each smoke constituent (1). The delivery of each of the organic chemicals found in MTS (e.g., 1,3-butadiene, benzene, nicotine, and benzo[a]pyrene) thus depends in a fundamental manner on the gas/particle (G/P) partitioning process. For a volatile MTS constituent such as 1,3-butadiene, the fraction fg that is in the gas phase will be close to 1. † ‡

OHSU Cancer Institute. OGI School of Science & Engineering.

If a volatile constituent of inhaled MTS is also of high to moderate water solubility, deposition will occur rapidly within the upper RT. In contrast, for a very low volatility compound (e.g., benzo[a]pyrene), the vast majority of the compound will be in the particle phase, and so, the compound will deposit only as fast and where the inhaled particles are deposited. We are interested here in an improved understanding of the compound dependence of G/P partitioning in MTS from commercial cigarettes. For each compound of interest, the G/P partitioning constant Kp (m3 µg-1) in a smoke sample of interest is given by (1, 2)

Kp )

cp cg

(1)

where cp (ng µg-1) is the concentration in the particle phase and cg (ng m-3) is the concentration in the gas phase. Because the particles occupy only a small fraction of the smoke volume, the units for cg (ng m-3) are

10.1021/tx0342316 CCC: $27.50 © 2004 American Chemical Society Published on Web 05/11/2004

806

Chem. Res. Toxicol., Vol. 17, No. 6, 2004

Pankow et al.

essentially equivalent to ng of compound per m3 of smoke. At a given total PM concentration (TPM, µg per m3 of smoke), the product cpTPM gives the particle-associated concentration in the smoke in units of ng per m3 of smoke. Following Pankow et al. (1-3), we then obtain eqs 2-5:

amount in particle phase ) (mass mass amount in gas phase )

cpTPM cg

Table 1. Equations Governing Compound-Dependent Kp Values for Tobacco Smoke PM (1, 2)

(2)

At equilibrium, by eq 1

amount in particle phase ) K TPM (mass mass amount in gas phase ) e

p

(3)

The equilibrium fraction in the gas phase is then

fg,e )

1 1 + KpTPM

(4)

and the equilibrium fraction in the particle phase is

fp,e )

KpTPM 1 + KpTPM

(5)

By definition, fg,e + fp,e ) 1. It is the magnitude of KpTPM as compared to 1 that determines which of the two fractions is larger at equilibrium. When KpTPM < 1, then fg,e > 0.5; when KpTPM > 1, then fp,e > 0.5. For partitioning to aerosol PM such as that found in tobacco smoke, theory (2) predicts that it is the volatility of a compound as represented by its vapor pressure in pure liquid form (poL, Torr) that plays the primary role in determining the compound dependency of Kp (see Table 1). Highly volatile compounds are thus expected to have low Kp values and so tend to exist mostly in the gas phase of tobacco smoke; lower volatility compounds exhibit relatively higher Kp values and so tend to exist predominantly in the particle phase of the smoke. Equation 7 in Table 1 describes the expected linear correlation between log Kp and log poL. Prior multiple compound studies of the dependence of log Kp on log poL values have included examinations of environmental tobacco smoke (ETS) (4, 5) and other types of aerosols (5-7) but not MTS. The available Kp values for MTS are limited to nicotine (1, 3, 8). Williamson and Allman (9) did determine fp values in MTS for 17 compounds ranging from acetaldehyde (poL ) 102.88 Torr at 20 °C) to toluene (poL ) 101.32 Torr at 20 °C). Figure 5 of their paper suggests a corresponding range for fp of ∼0.01 to ∼0.2. However, the fp values of Williamson and Allman (9) cannot be used to estimate experimental Kp values by assuming equilibrium conditions and then invoking eq 5 because TPM values for the MTS samples were not reported. The reported fp values are, nevertheless, quite consistent with what can be predicted by current theory. Specifically, one can probably assume that MWPM for conventional “full-flavored” cigarettes is about 100 g mol-1 (1) and that the activity coefficient ζ () solution phase “comfort factor”) values for the compounds of interest were on the order of ∼1 (1). For a volatility range of poL ) 102.88 to 101.32 Torr, then eq 6 predicts that the equivalent Kp range for acetaldehyde to toluene is ∼10-9.6 to ∼10-8.1 m3 µg-1 at 20 °C. Finally, if a mainstream TPM of ∼5 × 107 µg-1 m-3 is assumed, the fp range predicted using eq 5 for 20 °C is ∼0.01 to ∼0.3. [Note: TPM ) 5 × 107 µg-1 m-3 corresponds to 20

a ζ is evaluated on the mole fraction scale. On this scale, ζ values indicate how “comfortable” a compound feels being dissolved in a given solution phase relative to how it feels when it is present in its own pure liquid. When ζ ) 1, the compound feels equally comfortable; when 0 < ζ < 1, the compound is more comfortable; as ζ increases above 1, it feels increasingly uncomfortable.

mg of total PM () “tar” + particulate nicotine + particulate water) in ∼400 mL of smoke.] This study seeks to provide the first multiple compound examination for MTS of the dependence of smoke Kp values on compound volatility. The study compounds included a range of toxic and other VOCs found in MTS. Specific goals included determination of (i) cp and cg values for the study compounds in the MTS from two commercially relevant cigarette brands, (ii) the corresponding log Kp values, and (iii) the dependence of those log Kp values on log poL and temperature.

Experimental Procedures Cigarettes. Two brands of “full-flavored” cigarettes were purchased in cities in the U.S. market between October 2002 and July 2003. The pack relative humidity (RH, %) at ∼20 °C was measured on the day that a brand was smoked as follows: Marlboro filtered King Size (hard pack) RH ) 68%; Camel filtered King Size (hard pack) RH ) 64%. The cigarettes were not equilibrated to a single RH under the assumption that cigarette water content is a specific design parameter. Method. Each MTS sample was collected by smoking two cigarettes into a Teflon bag at 20 °C. The apparatus and sampling procedure used were the same as described by Pankow et al. (8) except that all of the puffs were drawn into a single bag. The procedure involved the simultaneous smoking of two cigarettes using 90 mL, 2 s puffs (45 mL per cigarette) at a puff interval of 30 s. Compound-dependent Kp values for the PM from

VOCs in Mainstream Tobacco Smoke

Chem. Res. Toxicol., Vol. 17, No. 6, 2004 807

the MTS were determined as follows. Fifteen minutes after collecting V1 ) 0.90 × 10-3 m3 of MTS containing mPM (µg) of PM, gas samples were taken for determination of the initial equilibrium gas phase concentration cg,1 (ng m-3). The bag was then emptied of its gaseous contents (essentially all of the PM had by that time collected on the walls). The bag was refilled with V2 ) 0.30 × 10-3 m3 of humidified nitrogen (RH ) 100%), and the collected PM was equilibrated with that new volume. [Numerous reported measurements indicate RH values near 100% in the gas phase of MTS from a variety of types of cigarettes, including “full-flavored” (10, 11) as well as “light” and “ultralight” (11) cigarettes.] Gas samples were then taken for the determination of the new equilibrium concentration cg,2 (ng m-3). The value of mPM was determined gravimetrically at the end of the experiment thereby providing

TPM1 ) mPM/V1

(8)

and

TPM2 ) mPM/V2

(10)

where m1 (ng) and m2 (ng) are the masses of the compound of interest within the collected PM phase at equilibrium with gas volumes V1 and V2, respectively. Thus,

cp,1 ) m1/mPM

(11)

cp,2 ) m2/mPM

(12)

It is volatilization of the compound from the collected PM into V2 that reduces m1 to m2 according to

m2 ) m1 - cg,2V2

(13)

cg,1 m1 ) m2 cg,2

(14)

By eq 10,

so that

m2 )

(cg,2)2V2 cg,1 - cg,2

(15)

Kp values were thus determined as

Kp )

cg,2V2 (cg,1 - cg,2)mPM

(16)

or equivalently

Kp )

V2 (cg,1/cg,2 - 1)mPM

mtot ) m1 + cg,1V1

)

(17)

(18)

cg,1 + cg,1V1 cg,2

(19)

cg,1cg,2V2 + cg,1V1 cg,1 - cg,2

(20)

) m2

and

[

) cg,1

]

cg,2V2 + V1 cg,1 - cg,2

(21)

For fg,e and fp,e, we have

fg,e )

cg,1V1 cg,1V1 + m1

(22)

fp,e )

m1 cg,1V1 + m1

(23)

(9)

The overall volatility of tobacco smoke PM is sufficiently low that mPM for a given tobacco smoke sample is not significantly reduced by the process of emptying and then refilling the collection bag. Each gas sample (5-10 mL) was drawn through two 0.5 µm pore size Teflon membrane filters in series and then through “air toxic” sorbent cartridges (Perkin-Elmer, Wellesley, MA) containing the sorbents Carbotrap and Carboxen 1000. Method Equations. By eq 1,

m1/mPM m2/mPM Kp ) ) cg,1 cg,2

The total mass mtot (ng) of a given compound in the initial MTS sample is given by

Consideration of the Possible Effects of Particle Agglomeration on Smoke PM Chemistry. Conversion of the liquid smoke PM from a suspension of submicrometer particles to a layer of liquid on the walls of a collection bag can alter the poL values of compounds in the PM through the Kelvin effect (12), which is described by o pL,r o pL,∞

)

o pL,r

poL

) exp

2σV h [rRT ]

(24)

o is the pure compound vapor pressure for a The parameter pL,r o () poL) is the vapor pressure of the bulk particle of radius r, pL,∞ liquid material (i.e., when r ) ∞), σ is the surface tension of the liquid, and V h is the molar volume of the compound. Available evidence (e.g., 13) suggests that most of the PM in conventional MTS is made up of particles with r > 0.1 µm () 10-5 cm). For o σ ≈ 50 dyne cm-1, V h ≈ 50 cm3 mol-1, eq 24 gives pL,r /poL ) 1.02, for which the effect of particle agglomeration on compound volatility will be small (∼2% or less). These calculations therefore indicate that compound-dependent volatility, which is the quantity of interest in these measurements, will not be significantly affected by conversion of r g 0.1 µm particles to a film on the walls of a collection bag. Chemical Analyses. In the determination of the cg,1 and cg,2 values, the air toxic sorbent cartridges were analyzed by thermal desorption using the procedures and instruments described by Pankow et al. (14). Prior to sampling, each cartridge was loaded with three internal standard compounds at 40 ng each: fluorobenzene, 4-bromofluorobenzene, and 1,4-dichlorobenzene-d4. Standard cartridges contained the same internal standard compounds plus levels of the target analyte compounds that ranged from 2 to 200 ng. The ability of the cartridges to quantitatively collect and release the gas phase analytes was verified by means of spike/recovery and breakthrough experiments as described elsewhere (14). Volatility Range for Application of the Method. The measurement of Kp values by application of eq 16 (or 17) requires reliable measurement of the quantity (cg,1/cg,2 - 1). This requirement establishes a specific range in compound volatility over which the method developed here can be used, and that range will depend on TPM2 as given by eq 9. Specifically, the compound may be neither (i) so highly volatile that cg,2 is extremely low and difficult to determine nor (ii) so nonvolatile that little volatilization occurs, making cp,2 ≈ cp,1 so that cg,2 ≈ cg,1, and therefore that cg,1/cg,2 difficult to distinguish from 1. The magnitude of typical analytical errors (e.g., (10%) means that (i) at the high volatility end, we need (cg,1/cg,2 - 1) < ∼10;

808

Chem. Res. Toxicol., Vol. 17, No. 6, 2004

Pankow et al.

Table 2. Properties of Selected VOCs, Nicotine, PAHs, and n-Alkanes

compound

log poL 20 °C (Torr)a

log poL 37 °C (Torr)a

poL(293.15) × 310.15

VOCs 54.09 76.14 58.08 74.08 53.06 72.11 78.11 92.14 100.16 106.16 106.16 106.16 104.15 120.19 120.19 120.19 120.19 120.19 134.22 120.20 134.22 134.22 134.22 134.22 134.22 128.17

3.26 2.47 2.27 2.23 1.94 1.88 1.88 1.34 0.90 0.86 0.81 0.69 0.65 0.52 0.40 0.28 0.25 0.22 0.09 0.09 0.01 -0.15 -0.41 -0.46 -0.58 -0.74

3.48 2.75 2.58 2.55 2.26 2.21 2.21 1.71 1.32 1.27 1.23 1.12 1.09 0.97 0.85 0.73 0.71 0.67 0.58 0.55 0.51 0.36 0.09 0.06 -0.06 -0.21

0.64 0.56 0.52 0.51 0.51 0.49 0.49 0.45 0.40 0.41 0.40 0.40 0.39 0.38 0.37 0.37 0.36 0.37 0.35 0.37 0.34 0.33 0.33 0.32 0.32 0.31

alkaloids 162.23

-1.61

-1.01

0.26

-2.71 -3.50 -3.53 -4.53 -4.73 -6.05

-2.04 -2.75 -2.78 -3.70 -3.88 -5.08

0.22 0.19 0.19 0.16 0.15 0.11

-3.04 -3.55 -4.07 -4.58 -5.09 -5.60

-2.28 -2.75 -3.22 -3.69 -4.16 -4.63

0.19 0.17 0.15 0.14 0.12 0.11

MW (g mol-1)

formula

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

1,3-butadiene carbon disulfide acetone methyl acetate acrylonitrile 2-butanone benzene toluene 2-hexanone ethylbenzene m,p-xylene o-xylene styrene isopropylbenzene n-propylbenzene 2-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene sec-butylbenzene 1,2,3-trimethylbenzene 1-isopropyl-4-methylbenzene n-butylbenzene 1,2,4,5-tetramethylbenzene 1,2,3,5-tetramethylbenzene 1,2,3,4-tetramethylbenzene naphthalene

C4H6 CS2 C3H6O C3H6O2 C3H3N C4H8O C6H6 C7H8 C6H12O C8H10 C8H10 C8H10 C8H8 C9H12 C9H12 C9H12 C9H12 C9H12 C10H14 C9H12 C10H14 C10H14 C10H14 C10H14 C10H14 C10H8

27

nicotine (as free-base)

C10H14N2

28 29 30 31 32 33

fluorene phenanthrene anthracene fluoranthene pyrene chrysene

C13H10 C14H10 C14H10 C16H10 C16H10 C18H12

PAHs 166.22 178.23 178.23 202.25 202.25 228.29

34 35 36 37 38 39

C16 C17 C18 C19 C20 C21

C16H34 C17H36 C18H38 C19H40 C20H42 C21H44

n-alkanes 226.44 240.47 254.49 268.52 282.55 296.57

poL(310.15) × 293.15

a Vapor pressure relations with temperature were obtained as follows: VOCs, ref 15; nicotine, ref 16 (as cited in ref 17); PAHs, ref 18; and n-alkanes, ref 19.

and (ii) at the low volatility end, we need (cg,1/cg,2 - 1) > ∼0.1. The method can thus be applied for compounds and smoke conditions that satisfy

10 > (cg,1/cg,2 - 1) > 0.1

(25)

The measurements discussed below suggest that for the VOCs and MTS PM studied here, the activity coefficient ζ () solution phase comfort factor) ranged from ∼1 to ∼16. The geometric midpoint of this range is ζ ≈ 4. If MWPM ≈ 100 g mol-1 (see ref 1), then by eq 6, we have Kp ≈ 4.6 × 10-8/poL ≈ 10-7.3/poL and so eq 28 becomes

By eq 17, this becomes

0.1 < KpmPM/V2 < 10

(26)

0.1 < KpTPM2 < 10

(27)

∼102.4 > poL > ∼10-1.6 Torr

(TPM1 ≈ 5 × 107 µg m-3,

MWPM ≈ 100 g mol-1, and ζ ≈ 4) (29)

or by eq 9,

Results and Discussion

The range of values for V2 that are experimentally easily accessible extends perhaps from ∼0.1V1 up to ∼10V1. The corresponding range of easily accessible values is TPM2 ≈ 10TPM1 down to ∼0.1TPM1. For a cigarette delivering ∼20 mg of total PM (“tar” + nicotine + water) in ∼400 mL of smoke, TPM1 corresponds to ∼5 × 107 µg m-3, and so, eq 27 gives the range of easily accessible Kp values as approximately

10-9.7 < Kp < 10-5.7

(for TPM1 ≈ 5 × 107 µg m-3)

(28)

Camel vs Marlboro: Correlation of Log Kp Values. Table 2 provides selected properties of the 26 VOCs studied here, nicotine, and the polycyclic aromatic hydrocarbons (PAHs) and n-alkanes for which Pankow et al. (4) have measured Kp values in ETS. Table 3 summarizes the VOC data obtained here for the Camel brand; Table 4 summarizes the data for Marlboro. Figure 1 is a plot of log Kp-Camel vs log Kp-Marlboro for the VOCs. Also plotted in Figure 1 is a point, derived from Pankow et al. (8), for nicotine in the neutral free-base (fb) form;

VOCs in Mainstream Tobacco Smoke

Chem. Res. Toxicol., Vol. 17, No. 6, 2004 809

Table 3. Results for MTS from Camel Cigarettes sample 1: mPM ) 3.30 × 104 µg cig-1 (sample volume: 450 mL cig-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

sample 2: mPM ) 2.43 × 104 µg cig-1 (sample volume: 450 mL cig-1)

compound

log cg,1 (ng m-3)

log cg,2 (ng m-3)

log mtot (ng cig-1)

log Kp (m3 µg-1)

log cg,1 (ng m-3)

log cg,2 (ng m-3)

log mtot (ng cig-1)

log Kp (m3 µg-1)

1,3-butadiene carbon disulfide acetone methyl acetate acrylonitrile 2-butanone benzene toluene 2-hexanone ethylbenzene m,p-xylene o-xylene styrene isopropylbenzene n-propylbenzene 2-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene sec-butylbenzene 1,2,3-trimethylbenzene 1-isopropyl-4-methylbenzene n-butylbenzene 1,2,4,5-tetramethylbenzene 1,2,3,5-tetramethylbenzene 1,2,3,4-tetramethylbenzene naphthalene

7.98 6.75 9.15 7.98 7.70 8.55 8.13 8.49 6.93 7.37 7.71 6.97 7.37 5.91 6.47 6.28 6.20 6.58 5.11 6.57 5.87 6.12 5.41 5.66 5.44 6.01

6.27 5.20 8.06 6.70 6.55 7.53 7.09 7.76 6.52 6.69 7.11 6.36 6.81 5.34 5.93 5.91 5.77 6.29 4.96 6.45 5.75 6.06 5.37 5.60 5.35 5.98

4.64 3.41 5.82 4.65 4.36 5.22 4.79 5.17 3.67 4.06 4.40 3.67 4.08 2.61 3.18 3.03 2.93 3.36 2.02 3.54 2.83 3.26 2.76 2.82 2.47 3.54

-10.05 -9.88 -9.40 -9.61 -9.46 -9.32 -9.34 -8.98 -8.53 -8.92 -8.82 -8.82 -8.76 -8.77 -8.74 -8.49 -8.58 -8.31 -7.97 -7.84 -7.84 -7.56 -7.26 -7.52 -7.73 -7.04

7.98 6.96 9.20 7.83 7.71 8.51 8.06 8.34 6.82 7.49 7.70 6.99 7.40 5.92 6.45 6.32 6.22 6.66 5.20 6.66 5.95 6.24 5.44 5.72 5.51 6.11

6.28 5.44 8.07 6.64 6.61 7.53 7.16 7.49 6.52 7.03 7.27 6.69 7.17 5.67 6.24 6.13 6.07 6.50 5.06 6.48 5.73 6.05 5.36 5.67 5.41 6.09

4.64 3.62 5.87 4.49 4.37 5.18 4.73 5.02 3.60 4.21 4.43 3.77 4.22 2.73 3.29 3.18 3.13 3.56 2.12 3.53 2.78 3.10 2.52 2.91 2.54 3.70

-9.91 -9.72 -9.31 -9.36 -9.27 -9.14 -9.05 -8.99 -8.21 -8.49 -8.44 -8.19 -8.05 -8.10 -8.02 -7.93 -7.84 -7.84 -7.81 -7.94 -8.01 -7.94 -7.51 -7.36 -7.60 -6.85

Table 4. Results for MTS from Marlboro Cigarettes sample 1: mPM ) 2.26 × 104 µg (sample volume: 450 mL cig-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

sample 2: mPM ) 1.80 × 104 µg cig-1 (sample volume: 450 mL cig-1)

compound

log cg,1 (ng m-3)

log cg,2 (ng m-3)

log mtot (ng cig-1)

log Kp (m3 µg-1)

log cg,1 (ng m-3)

log cg,2 (ng m-3)

log mtot (ng cig-1)

log Kp (m3 µg-1)

1,3-butadiene carbon disulfide acetone methyl acetate acrylonitrile 2-butanone benzene toluene 2-hexanone ethylbenzene m,p-xylene o-xylene styrene isopropylbenzene n-propylbenzene 2-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene sec-butylbenzene 1,2,3-trimethylbenzene 1-isopropyl-4-methylbenzene n-butylbenzene 1,2,4,5-tetramethylbenzene 1,2,3,5-tetramethylbenzene 1,2,3,4-tetramethylbenzene naphthalene

8.33 6.99 9.26 7.96 8.01 8.59 8.00 8.42 7.19 7.23 7.53 6.81 7.20 5.80 6.35 6.14 6.08 6.52 5.20 6.56 5.91 6.05 5.35 5.66 5.48 5.82

6.54 5.43 7.60 6.63 6.70 7.47 6.86 7.61 6.73 6.80 7.14 6.49 6.92 5.51 6.07 5.95 5.88 6.37 5.08 6.31 5.63 5.79 5.25 5.49 5.33 5.79

4.99 3.64 5.91 4.62 4.67 5.25 4.66 5.10 3.91 3.96 4.27 3.58 3.99 2.59 3.14 3.00 2.93 3.43 2.17 3.37 2.70 2.85 2.36 2.54 2.39 3.22

-9.96 -9.72 -9.83 -9.48 -9.46 -9.25 -9.28 -8.91 -8.46 -8.41 -8.33 -8.22 -8.14 -8.15 -8.12 -7.90 -7.93 -7.80 -7.66 -8.06 -8.13 -8.08 -7.60 -7.86 -7.79 -7.04

8.27 6.81 9.12 7.88 7.79 8.42 8.08 8.29 7.11 7.24 7.50 6.87 7.23 5.84 6.37 6.18 6.09 6.55 5.20 6.53 5.89 6.08 5.32 5.63 5.39 5.85

6.41 5.31 7.47 6.41 6.25 7.26 6.95 7.50 6.22 6.71 7.12 6.46 6.90 5.42 6.07 6.02 5.97 6.46 4.95 6.35 5.65 5.88 5.28 5.60 5.36 5.81

4.93 3.47 5.77 4.54 4.45 5.08 4.74 4.97 3.83 3.95 4.25 3.61 4.00 2.58 3.15 3.07 3.05 3.60 2.01 3.41 2.70 2.92 2.64 3.00 2.79 3.20

-9.94 -9.57 -9.72 -9.53 -9.61 -9.21 -9.17 -8.80 -8.41 -8.45 -8.22 -8.28 -8.15 -8.29 -8.07 -7.73 -7.59 -7.44 -7.97 -7.77 -7.95 -7.87 -7.04 -6.98 -6.95 -7.00

for that point, log Kp,fb-Camel ) -4.99 and log Kp,fbMarlboro ) -5.12. The data points in Figure 1 are all closely distributed around the 1:1 line. This result supports the suggestion by Pankow (1) that similar brands of cigarettes tend to produce MTS PM that is roughly similar with respect to the major compositional properties that determine the G/P partitioning of neutral organic compounds. The most important of these properties are (i) MWPM and (ii) overall phase polarity. The role of MWPM is described explicitly in eq 6. The role of

the overall phase polarity is represented implicitly in eq 6 through its effects on the compound-dependent ζ values. On the basis of the good 1:1 correlation in Figure 1, we consider it likely that for the smoking conditions utilized here, MWPM-Camel ≈ MWPM-Marlboro, and that the overall phase polarities of the MTS PM from these two brands of “full-flavored” cigarettes were similar. The VOC data were therefore pooled by computing the arithmetic means of all available replicate Kp deter-

810

Chem. Res. Toxicol., Vol. 17, No. 6, 2004

Pankow et al. Table 5. Values of log K h p (m3 µg-1) at 20 °C and ζ for VOCs in MTS, Nicotine in MTS, Selected PAHs in ETS, and Selected n-Alkanes in ETS log K hp log Kp (m3 µg-1) (m3 µg-1) at 20 °C at 37 °C ζ at 20 °C (measured)a (estimated)b (estimated)c

Figure 1. Log Kp-Camel vs log Kp-Marlboro for partitioning of VOCs and free-base nicotine to MTS PM from Camel and Marlboro cigarettes at 20 °C.

minations [two for Camel (from Table 3) and two for Marlboro (from Table 4)]. Table 5 summarizes the resulting log K h p values for the 26 VOCs. Table 5 also includes log K h p values for free-base nicotine in MTS as computed here based on data from Pankow et al. (8) and log K hp values for selected PAHs and n-alkanes from Pankow et al. (4). Correlations between Log K h p and log poL. Figure 2 provides a plot of log K h p vs log poL for the Table 5 MTS data for the 26 VOCs. The data point for nicotine is also included. The 26 VOC data are highly correlated (m ) -0.77, b ) -7.76, and r ) -0.99). The fact that the slope m for the 26 VOCs is different from -1.0 by an amount that is statistically significant may be understood as follows. The only compound-dependent parameters in eq 6 are poL and ζ (MWPM depends only on the chemical composition of the PM). If ζ was the same for all 26 compounds, the slope m would equal -1.0 exactly. The fact that m for the 26 VOCs is less steep than -1.0 indicates a trend by which the compound-dependent ζ values in the PM tend to increase as log poL decreases. The log K h p values thereby increase less rapidly as log poL decreases than would have been the case if ζ ≈ constant. The 26 VOCs span 4 orders of magnitude in volatility (log poL ) 3.26 at 20 °C down to log poL ) -0.74 at 20 °C). Over that range, to obtain a slope of -0.77, the factor increase in ζ can be estimated as ∼10[(1-0.77)(4)] ≈ 10. The PM phase ζ values for the 26 VOCs were estimated (see Table 5) using (i) eq 6, (ii) the K h p values at 20 °C from this study, (iii) the poL values at 20 °C in Table 2, and (iv) MWPM ) 100 g mol-1 (see ref 1). As expected, the PM phase ζ values in Table 5 increase by about a factor of 10 going from 1,3-butadiene to naphthalene. Origins of the Compound Dependence of the Activity Coefficient ζ in Smoke PM. For the 26 VOCs, the trend of increasing the ζ values (Table 5) with decreasing the poL is likely the result of the moderate polarity of MTS PM from conventional “full-flavored” cigarettes. This polarity is caused by the presence in such PM of numerous nitrogen- and oxygen-containing compounds, including substantial nicotine and water. Indeed, when dissolved in a moderately polar phase, a series of compounds of increasing hydrophobicity, e.g., the group

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

VOCs in MTS 1,3-butadiene carbon disulfide acetone methyl acetate acrylonitrile 2-butanone benzene toluene 2-hexanone ethylbenzene m,p-xylene o-xylene styrene isopropylbenzene n-propylbenzene 2-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene sec-butylbenzene 1,2,3-trimethylbenzene 1-isopropyl-4-methylbenzene n-butylbenzene 1,2,4,5-tetramethylbenzene 1,2,3,5-tetramethylbenzene 1,2,3,4-tetramethylbenzene naphthalene

-9.96 -9.71 -9.51 -9.49 -9.43 -9.23 -9.19 -8.91 -8.38 -8.53 -8.40 -8.32 -8.21 -8.26 -8.16 -7.94 -7.87 -7.75 -7.83 -7.89 -7.97

0.93 3.2 3.2 3.3 5.7 4.1 3.8 6.9 5.6 8.6 7.2 7.7 6.6 10.1 10.7 8.4 7.5 6.2 10.0 11.6 16.5

-10.16 -9.96 -9.80 -9.78 -9.73 -9.54 -9.50 -9.26 -8.78 -8.91 -8.80 -8.72 -8.62 -8.69 -8.59 -8.37 -8.30 -8.18 -8.29 -8.32 -8.44

-7.82 -7.30

16.8 9.4

-8.30 -7.78

-7.32

11.0

-7.81

-7.37

16.1

-7.86

-6.97

9.4

-7.48

alkaloids in MTSd 27 nicotine (as free-base)

-5.05

0.80

-5.63

28 29 30 31 32 33

PAHs in ETSe fluorene phenanthrene anthracene fluoranthene pyrene chrysene

-4.89 -3.44 -3.13 -2.51 -2.34 -1.26

7.3 1.6 0.83 2.0 2.1 3.7

-5.54 -4.16 -3.85 -3.32 -3.16 -2.21

34 35 36 37 38 39

n-alkanes in ETSe C16 C17 C18 C19 C20 C21

-4.70 -4.11 -3.70 -3.22 -2.73 -2.23

26

10.0 11.6 15.2 17.3 18.4 19.6

-5.44 -4.89 -4.52 -4.09 -3.64 -3.18

a Averages for Camel (N ) 2) and Marlboro (N ) 2) data obtained in this study. b ζ values at 20 °C calculated by eq 6 using K h p values at 20 °C from this study, poL values at 20 °C (Table 2), and assuming MWPM ) 100 g mol-1. c Kp values at 37 °C calculated based on K h p values at 20 °C from this study and eq 31. d K hp value for free-base nicotine at 20 °C based on averaging of Camel and Marlboro data in Pankow et al. (8). e K h p values for PAHs and n-alkanes at 20 °C are based on data at 25 °C in Pankow et al. (4), after correction to 20 °C by eq 31.

benzene, toluene, ethylbenzene, n-propyl benzene, and n-butyl benzene, will exhibit ζ values that increase over the series. From Table 5, the estimated ζ values (20 °C) for this group in the MTS PM studied here are 3.8, 6.9, 8.6, 10.7, and 16.8, respectively. By way of comparison, when these compounds are dissolved at 20 °C in a completely aqueous phase (which is quite polar), their ζ values can be calculated (by consideration of their solubilities in water) to increase according to 103.4, 104.0, 104.6, 105.1, and 105.7, respectively. On the other hand, when dissolved in n-hexane (which is completely nonpolar), all of these compounds exhibit ζ values that are

VOCs in Mainstream Tobacco Smoke

Chem. Res. Toxicol., Vol. 17, No. 6, 2004 811 Table 6. Values of log K h p (m3 µg-1), log poL (Torr), and ζ (All at 20 °C) for Volatile n-Alkanes in MTS

n-alkanes in MTS

log K hp at 20 °C (m3 µg-1) (measured)a

log poL at 20 °C (Torr)b

ζ at 20 °C (estimated)c

C6 (n-hexane) C7 (n-heptane) C8 (n-octane) C9 (n-nonane) C10 (n-decane) C11 (n-undecane)

-9.49 -9.19 -8.97 -8.56 -8.19 -7.64

2.09 1.56 0.99 0.41 -0.14 -0.58

4.60 7.86 17.3 25.8 39.5 30.4

a Averages for Camel (N ) 2) and Marlboro (N ) 2) data as obtained in this study. b Calculated based on Antoine parameters given in NIST Webbook (21). c ζ values at 20 °C calculated by eq 6 using K h p values at 20 °C from this study, poL values at 20 °C, and assuming MWPM ) 100 g mol-1.

Figure 2. Log K h p vs log poL for partitioning of VOCs and freebase nicotine to MTS PM from Camel and Marlboro cigarettes at 20 °C.

Figure 3. Log K h p vs log poL for partitioning of VOCs and freebase nicotine to MTS PM from Camel and Marlboro cigarettes at 20 °C, and log Kp vs log poL for partitioning of selected PAHs and n-alkanes to ETS PM at 20 °C.

very close to 1.0. The fact that ζ ≈ 1 for nicotine dissolved in MTS PM (see Table 5) is surely the consequence of nicotine’s remarkable amphiphilic nature; at ambient temperatures, nicotine is completely miscible with a wide range of both nonpolar and polar solvents, including water. The data points of Pankow et al. (4) for the partitioning of PAHs and n-alkanes to ETS PM lie above the extrapolated correlation line for partitioning of the 26 VOCs to MTS PM (Figure 3). The functionality of eq 6 indicates that Kp can be increased at a given poL only by decreasing the quantity (MWPMζ). Because ETS will not tend to exhibit a significantly lower MWPM than the MTS of the type studied here (low MW compounds tend to be lost from the PM phase during the formation/aging of ETS), the relative positions of the three data sets in Figure 3 suggest that the ζ values for the five PAHs and the six n-alkanes were smaller in the ETS PM studied by Pankow et al. (4) than they would have been found to be in the MTS PM studied here. For example, assuming for

the MTS that MWPM ≈ 100 g mol-1 (1), then for log poL ) -4.7 and 20 °C, the extrapolated 26 VOC correlation line yields ζ ≈ 130. In ETS, on the other hand, the data in Table 5 suggest for pyrene (log poL ) -4.7 at 20 °C) that ζ is about 2. Relatively lower ζ values for the PAHs and n-alkanes in ETS PM as compared to ζ values suggested by the 26 VOC correlation line for MTS PM could be due to (i) a greater affinity of the PAHs and n-alkanes for tobacco smoke PM relative to the 26 VOC correlation line and/or (ii) a lower polarity (and therefore more favorable solution environment for nonpolar organic compounds) in ETS PM relative to the MTS PM. The first possibility can probably be ruled out because data obtained here with six volatile n-alkanes using the same MTS samples (see Table 6) were found to lie somewhat below, and not above, the 26 VOC correlation line in Figure 3. Regarding the relative polarities of the ETS PM relative to the MTS PM, we note that ETS PM forms by the mixing/dilution of sidestream tobacco smoke and exhaled MTS. During that process, both nicotine and water will volatilize from the PM phase; the evaporation of such compounds probably leaves ETS PM less polar than MTS PM. Last, considering just the ETS data, the fact that the ETS PAHs line lies above the ETS n-alkanes line is likely due to (i) the retention of some significant polarity in the ETS PM and (ii) the fact that in a somewhat polar phase a group of PAHs spanning a certain poL range can be expected to exhibit lower ζ values (and therefore higher Kp values) than a group of n-alkanes of the same volatility range. The greater relative solution comfort (lower ζ) in polar phases of an aromatic compound vs an n-alkane of the same carbon number is a well-known consequence of dipole/induced dipole interactions between polar solvent molecules and the π electrons on aromatic compounds. For example, in water, the solubility of benzene is 10-1.64 mol L-1 while that of n-hexane at 25 °C is 10-3.89 mol L-1 (21). Effects of Temperature and TPM Levels on G/P Distributions. The functional character of eq 6 indicates that for a given type of smoke PM, for each compound of interest

Kp ∝

T poL

(30)

Assuming that ζ values are roughly independent of temperature (certainly as compared to the strong temperature dependence of poL values), for a given sample of

812

Chem. Res. Toxicol., Vol. 17, No. 6, 2004

Pankow et al.

where Kp(T2) and poL(T2) are the Kp and poL values at T2 (K) and Kp(T1) and poL(T1) are similarly defined for T1 (K). Values of the factor poL(293.15) × 310.15[poL(310.15) × 293.15] are given in Table 1 for adjustment of Kp values at 293.15 K (20 °C) to the corresponding Kp values at the normal human temperature of 310.15 K (37 °C). It is wellknown that poL values increase strongly with temperature and that this dependence is much stronger than the temperature dependence of T (K) itself. The result is that Kp values decrease strongly with increasing temperature (see the approximation in eq 32). The temperature correction factors in Table 1 were used to adjust the MTS Kp values obtained here for the VOCs at 293.15 (20 °C) to 310.15 K (37 °C); the results are given in Table 5. Figure 4a-c provide plots of fg,e vs compound number for three combinations of temperature and TPM. During the initial stages of smoking a cigarette, the MTS can exit the cigarette at near-ambient temperatures. Figure 4a thus pertains to a temperature of 20 °C and a typical MTS TPM value of ∼5 × 107 µg m-3; the associated fg,e values extend from 0.16 for naphthalene (compound 26) up to 0.995 for 1,3-butadiene (compound 1). Inhaling MTS can raise the temperature of the smoke from ambient levels to 37 °C and will also lower the TPM level due to dilution. Both of these changes will cause fg,e values to increase; Figure 4b illustrates the effects of raising T to 37 °C while maintaining TPM ) 5 × 107 µg m-3; Figure 4c illustrates the combined effects of raising T to 37 °C and lowering TPM by a factor of 10 to TPM ) 5 × 106 µg m-3.

Conclusions

Figure 4. Fraction in the gas phase of MTS at equilibrium (fg,e) vs VOC compound number for naphthalene (compound 26) to 1,3-butadiene (compound 1) for various conditions. Conditions: (a) 20 °C and TPM ) 5 × 107 µg m-3; (b) 37 °C and TPM ) 5 × 107 µg m-3; and (c) 37 °C and TPM ) 5 × 106 µg m-3.

smoke PM, the temperature dependence of Kp of a given compound may be estimated according to

Kp(T2) ) Kp(T1) ×

poL(T1) × T2 poL(T2) × T1

(31)

References

which may be approximated as

Kp(T2) ≈ Kp(T1) ×

poL(T1) poL(T2)

The vapor pressure-dependent G/P partitioning behavior of 26 VOCs of toxicological interest in MTS has been found to closely follow the theory of Pankow (2). The utility of this unifying result in predictions of chemical behavior in tobacco smoke is augmented by the finding that two different brands of domestic “full-flavored” cigarettes yielded smoke PM with very similar G/P sorption properties for neutral organic compounds. The significant polarity of MTS PM affects the activity coefficient (ζ) values of partitioning compounds in the PM phase so that (i) for a group of compounds, a trend of increasing ζ (i.e., increasing discomfort when dissolved in the PM phase) with decreasing poL causes the corresponding log Kp values to increase less strongly with decreasing log poL than if the ζ values were all equal; (ii) for two compounds of equal poL, the compound with the larger ζ will exhibit the lower Kp; and (iii) for a compound with moderate to significant hydrophobicity, its Kp will likely be larger for partitioning to the PM of ETS than for partitioning to the relatively more polar PM of MTS. For a given smoke sample (including inhaled smoke), the compound-dependent gas phase fraction at equilibrium (fg,e) increases strongly with increasing temperature and decreasing total PM concentration (µg m-3).

(32)

(1) Pankow, J. F. (2001) A consideration of the role of gas/particle partitioning in the deposition of nicotine and other tobacco smoke compounds in the respiratory tract. Chem. Res. Toxicol. 14, 14651481. (2) Pankow, J. F. (1994) An absorption model of gas/particle partitioning in the atmosphere. Atmos. Environ. 28, 185-188.

VOCs in Mainstream Tobacco Smoke (3) Pankow, J. F., Mader, B. T., Isabelle, L. M., Luo, W., Pavlick, A., and Liang, C. (1997) Conversion of nicotine in tobacco smoke to its volatile and available free-base form through the action of gaseous ammonia. Environ. Sci. Technol. 31, 2428-2433. See also Pankow, J. F. (1999) Environ. Sci. Technol. 33, 1320. (4) Pankow, J. F., Isabelle, L. M., Buchholz, D. A., Luo, W., and Reeves, B. D. (1994) Gas/particle partitioning of polycyclic aromatic hydrocarbons and other compounds to environmental tobacco smoke. Environ. Sci. Technol. 28, 363-365. (5) Liang, C., and Pankow, J. F. (1996) Gas/particle partitioning of organic compounds to environmental tobacco smoke: partition coefficient measurements by desorption and comparison to urban particulate material. Environ. Sci. Technol. 30, 2800-2805. (6) Liang, C., Pankow, J. F., Odum, J. R., and Seinfeld, J. H. (1997) Gas/particle partitioning of semivolatile organic compounds to model inorganic, organic, and ambient smog aerosols. Environ. Sci. Technol. 31, 3086-3092. (7) Lohmann, R., Harner, T., Thomas, G. O., and Jones, K. C. (2000) A comparative study of the gas-particle partitioning of PCDD/ Fs, PCBs, and PAHs. Environ. Sci. Technol. 34, 4943-4951. (8) Pankow, J. F., Tavakoli, A. D., Luo, W., and Isabelle, L. M. (2003) Percent free-base nicotine in the tobacco smoke particulate matter of selected commercial and reference cigarettes. Chem. Res. Toxicol. 16, 1014-1018. (9) Williamson, J. T., and Allman, D. R. (1966) The distribution of tobacco smoke constituents between the vapor and particle phases. Beitrage Tabakforshung Int. (Contrib. Tob. Res.) 3, 590596. Currently available at http://www.beitraege-bti.de/nxt/ gateway.dll?f)templates&fn)default.htm&vid)btfi: btfi. (10) Thome, F. A. (1965) Gas chromatographic determination of the water in cigarette mainstream smoke and total particulate matter. Report at the 19th Tobacco Chemists’ Research Conference, Lexington, Kentucky, October 26-28, 1965.

Chem. Res. Toxicol., Vol. 17, No. 6, 2004 813 (11) Lauterbach, R. H. (2001) Application of gas-particle partitioning theory to water in cigarette mainstream smoke. Presentation at the 55th Tobacco Science Research Conference, Greensboro, North Carolina, Sept. 9-12, 2001. (12) Seinfeld, J. H., and Pandis, S. N. (1998) Atmospheric Chemistry and Physics, p 1326, Wiley, New York. (13) Davies, C. N. (1988) Cigarette smoke: generation and properties of the aerosol. J. Aerosol Sci. 19, 463-469. (14) Pankow, J. F., Luo, W., Isabelle, L. M., Bender, D. A., and Baker, R. J. (1998) Determination of a wide range of volatile organic compounds in ambient air using multisorbent adsorption/thermal desorption and gas chromatography/mass spectrometry. Anal. Chem. 70, 5213-5221. (15) Yaws, C. L. (1999) Vapor pressuresOrganic compounds. Chemical Properties Handbook, McGraw-Hill, New York. (16) Frost, B. E. Measurement of the Vapor Pressure of Nicotine, internal report, report no. C.33/P/86/NR300, Rothmans International Services Ltd. (17) Lewis, D. A. (1994) Ph.D. Thesis, Department of Chemistry and Biological Chemistry, University of Essex, U.K. (18) Yamasaki, H., Kuwata, K., and Miyamoto, K. (1984) Determination of the vapor pressure of polycyclic aromatic hydrocarbons in the supercooled liquid phase and their adsorption on airborne particulate matter. Chem. Soc. Jpn. 8, 1324-1329. (19) Makar, P. A. (2001) The estimation of organic gas vapour pressure. Atmos. Environ. 35, 961-974. (20) Huibers, P. D. T., and Katritzky, A. R. (1998) Correlation of the aqueous solubility of hydrocarbons and halogenated hydrocarbons with molecular structure. J. Chem. Inf. Comput. Sci. 38, 283292. (21) National Institute of Standards and Technology, NIST Webbook, http://webbook.nist.gov/.

TX0342316