Gas-Particle Partitioning of Formaldehyde in Mainstream Cigarette Smoke

The experimental apparatus used as the basis for this denuder study has been described previously (25). We modified this apparatus slightly to optimize it for the study of aldehydes (Figure 2). In brief, the heart of the apparatus is the vertically positioned diffusion denuder tube made of thin-walled Pyrex®, with an internal diameter of 0.80 cm and length of 150.0 cm (our standard measurements) and coated internally with DNPH. The denuder tube temperature is regulated to within ± 0.2 K at each of the experimental temperatures of 298–323 K. The lower end of the denuder tube is attached to a Cambridge filter pad (CFP) for the collection of smoke particles leaving the denuder. The exit gases are passed through two impingers, filled with DNPH acetonitrile solution.

Figure 2

The air flows and the required volumetric flow rate (Q₂₇₃) are regulated and measured by electronic mass flow controllers, which along with the water-saturator, the gas mixer and the humidity sensor regulate and measure the RH of the sampling air. (The RH values of the sampling air that we report refer to RH values at the laboratory temperature, which was kept constant at 298 ± 2 K.)

This was conducted by applying multiple layers of DNPH from acetonitrile solution as detailed in the Supplementary Information.

Denuder experiments were carried out by continuously drawing aldehyde-air gas mixtures or cigarette MS through the denuder tube coated with DNPH, from which particles were collected on a CFP and two impingers in series (Figure 2). The CFP was impregnated with 0.25 g of DNPH, 10 mL of CH₃CN and 0.2 mL of 85% H₃PO₄ solution and the impingers were filled with an acidic solution of DNPH (20 mL of acetonitrile, 0.2 mL of phosphoric acid (85%) and 0.05 g of DNPH). Tip-ventilated Kentucky 3R4F reference cigarettes were used in the smoking experiments. Sample cigarettes were conditioned for 4–8 days at 298 K and 60% RH. Cigarettes were smoked vertically (Figure 2) using a continuous draw process, as described previously (25). Only the first 2 cm of each cigarette was sampled to minimize variations in the MS temperature and humidity when entering the denuder tube, consistent with a continuous sampling time of approximately 120 s.

After sampling, the denuder tube was cut into 5-cm sections. The aldehyde hydrazone derivatives deposited in each 5-cm section, and collected by the Cambridge filter and impingers, were determined by high-performance liquid chromatography (HPLC). Coatings from the 5-cm denuder segments were dissolved in 2 mL of CH₃CN and the Cambridge filter was extracted by 20 mL of CH₃CN. At least three replicates of denuder runs were made for each experimental condition (e.g., at least three cigarettes were smoked at each given temperature and RH and Q₂₇₃ value).

To measure aldehydes in DNPH hydrazone derivatives, an automated Agilent 1200 HPLC apparatus (Santa Clara, CA, USA) equipped with a Phenomenex LUNA C₁₈ column (Phenomenex, Macclesfield, UK); dimensions: particle size, 5 μm; length, 250 mm; internal diameter, 4.6 mm. The column was kept at 313 K; the flow rate of the mobile phase was 1.5 mL min⁻¹. Detection was achieved by ultraviolet absorption at λ = 360 nm. Calibrations were carried out with a TO11/IP-6A Aldehydes/Ketone-DNPH calibration standard (Sigma-Aldrich, Gillingham, UK) that contained hydrazones of 15 carbonyls in acetonitrile solution.

Two gradient elution methods were used for the analyses. The HPLC protocol described by Restek (40) was used for model aldehydes. Carbonyls in cigarette MS were analysed by a method developed by British American Tobacco (41). Further details of the HPLC analyses and representative chromatograms are presented as Supplementary Information.

Denuder depletion data for formaldehyde at different flow rates are presented in Figure 4. The rate of formaldehyde depletion along the denuder was significantly dependent on the flow rate of the formaldehyde-air flow; contact time and denuder efficiency increased with slowing air flow. The difference is substantial between the fastest and slowest air flows, but differences were smaller at slower flows.

Figure 4

The best denuder performance was found at Q₂₇₃ = 200 mL min⁻¹, but was not substantially different from that with Q₂₇₃ = 300 mL min⁻¹.

As noted above, denuder depletion slopes were faster at higher RHs (Figure 5), implying that the rate of sink wall reaction between formaldehyde and DNPH coating increases with higher water concentration in the gas stream. Increasing RH at low RH values strongly enhanced efficiency of denuder trapping, although differences between curves became smaller once RH reached 20%. This RH may be regarded as a lower limit for denuder determination of formaldehyde in smoking studies, since the burning process of the cigarette produces combustion water in the ingoing smoke stream.

Figure 5

Temperature had little to no effect on depletion of formaldehyde along the denuder tube (Figure 6). The denuder depletion curves seem to indicate a slight negative temperature dependence of the rate of the gas-solid reaction of formaldehyde with DNPH. However, the observation could also be explained by changes in the linear flow rate with temperature. Flow rate is lower at lower temperatures, which produces an increase in wall-gas contact time and a consequent higher wall-capture efficiency.

Figure 6

For our experimental conditions with formaldehyde, the Gormley-Kennedy [GK] equation is mD(z)/mD(z0)=0.819×exp(-7.314 h)=0.819×exp-7.314[z π DT/(2QT)] \eqalign{& {m_D}(z)/{m_D}({z_0}) = 0.819 \times \exp (- 7.314\,h) \cr & = 0.819 \times \exp - 7.314[z\,\pi \,{D_T}/(2{Q_T})] \cr} where m_D(z) is the formaldehyde mass collected in the j^th denuder section per μg; m_D(z₀) is the formaldehyde mass collected in an arbitrary 0^th denuder section (e.g., between 0 and 5 cm) per μg; z is the axial end point of the j^th denuder section from the entrance of the denuder tube per cm; D_T is the diffusion coefficient of formaldehyde in air at temperature T per cm² s⁻¹; and Q_T is the volumetric flow rate of air through the denuder at temperature T per mL s⁻¹. In the case of perfect denuder operation, the depletion of formaldehyde would follow the GK equation, and so a single parameter fit to the experimental data should return the gas-phase diffusion coefficient of formaldehyde as was found previously with nicotine (25). The diffusion coefficient of formaldehyde is well established as D₂₉₈ (formaldehyde, air) = 0.165 ± 0.020 cm² s⁻¹ (41, 42).

When the only fitting parameter was the diffusion coefficient of formaldehyde, the GK fitting procedure provided a very poor description of the experimental data (Table 2), giving an estimation of formaldehyde's diffusion coefficient around three to five times smaller than in the literature value. This might be because the formaldehyde mass is underestimated in the first segment of the denuder tube due to saturation effects and/or flow was not fully laminar in the entrance portion of the tube. With the use of two parameters (D₂₉₈ and the initial formaldehyde mass), formaldehyde decay data are described accurately (Table 2, Figure 4). The estimated D₂₉₈ values are consistent at flow rates below 400 mL min⁻¹ (Table 2), but lower at faster flow-rates. At the slowest flow the D₂₉₈ value of 0.184 cm² s⁻¹ is reasonably close to that in the literature (0.165 cm² s⁻¹), although the estimated initial formaldehyde masses do not agree with the determined mass balances. The denuder tube, therefore, is most likely not a ‘perfect sink’ even at our optimized operational conditions.

Similarly, for the effects of RH, the conclusions are very similar (but in the opposite direction) to those for flow rate (Table 2, Figure 5). For temperature, the single-parameter fits provide very low diffusion coefficients for formaldehyde, while with two-parameters, the estimated D₂₉₈ values decrease with increasing temperature (Table 2, Figure 6). This effect is more likely due to increasing linear flow rate with increasing temperatures. Nevertheless, the results clearly show that the dependence of the heterogeneous wall reaction on temperature is probably minor, which is advantageous for investigating G/P in cigarette smoke.

Based on the denuder behaviour of MS nicotine (25, 27, 29), double exponential depletion curves were anticipated for formaldehyde if ideal diffusion denuder operation was obtained and initial partitioning between phases was significant. Such a trend was observed with formaldehyde in 3R4F cigarette smoke (Figure 7). Importantly, deposition was much slower than found with formaldehyde vapour under the same conditions, where deposition was complete after formaldehyde traversed 20 cm of the denuder. With cigarette smoke, formaldehyde continued to deposit along the full 1.5 m length of the denuder, and a significant portion was trapped at the CFP and impinger placed after the denuder tube. This observation shows that formaldehyde in cigarette smoke is entrained within the aerosol to a significant degree, resulting in a significantly extended delivery than found with formaldehyde vapour alone.

Figure 7

Total masses of formaldehyde determined from denuder smoking experiments are presented in Table 3. Despite a directional decrease in total formaldehyde mass with increasing temperature, the three values are within 13% of the average value, indicating that the combustion processes of the cigarettes were not significantly influenced by the variation of the sampled air temperature in the denuder tube. As noted in the Introduction, to our knowledge, no denuder sampling study has been performed to determine aldehydes, but carbonyls are routinely quantified in cigarette smoke (37, 43,44,45). The values measured in our experiments are lower than those reported elsewhere as very different smoking protocols are used.

In our study, the need for laminar flow meant that we drew continuously on the cigarette, sampling only 2 cm of the 5.7-cm length of the burnt cigarette, whereas the other studies cited used higher flow rate puffing approaches, taken under the ISO or Health Canada Intense puffing regimes (46, 47) to generate smoke, by sampling the entire cigarette.

At 298, 310, and 323 K, formaldehyde displayed a relatively high deposition close to the denuder inlet, which declined sharply further along the tube, demonstrating dynamic changes in G/P partitioning (Figure 8). A small but definite increase of denuded formaldehyde mass was seen at all distances with increasing denuder temperature (Figure 8). In accordance, the fraction of formaldehyde deposited over the total length of the denuder tube was roughly 35% at 298 K and around 61% at 323 K (Table 3). At each temperature, deposition of formaldehyde from cigarette smoke occurred over a longer denuder tube distance than found with formaldehyde vapour alone. The formaldehyde depletion data have been analysed by the preferred Lipowicz-Piadé model (29).

Figure 8

In the current study, cigarette smoke formaldehyde concentrations showed dynamic changes along the denuder distance, similarly to those reported for nicotine (25). At room temperature, the initial gas-phase formaldehyde fraction of smoke obtained by the application of the theoretical model of Lipowicz and Piadé (29) is around 0.5%. This implies that the > 99% formaldehyde deposited in the denuder tubes originates of evaporation from smoke particles.

The corresponding parameter for nicotine is around 0.09%. Therefore, both formaldehyde and nicotine in mainstream cigarette smoke exist predominantly in the particle-phase, and most deposition in the denuder tube is owing to evaporation from particles, subsequent gas-phase diffusion, and wall adsorption. Our results are in line with the previously reported findings that significant portions of formaldehyde (10, 15, 16, 44) and nicotine are found in the particle-phase of cigarette MS (15).

The evaporation-diffusion-deposition process for formaldehyde is more rapid than that for nicotine, as shown by the measured total denuded formaldehyde fraction at room temperature, which is significantly higher than that determined for nicotine (roughly 35% versus 6%). Three explanations can be advanced to interpret this observation.

First, the evaporation rate of formaldehyde from cigarette MS particles is likely to exceed that of nicotine due to the significantly higher vapour pressure of formaldehyde (>10⁵ Pa) in comparison to nicotine (5.7 Pa) at room temperature.

The impact of temperature on G/P partitioning of the two species can be explained by use of the Lipowicz and Piadé (29) model, which predicts that the initial gas-phase formaldehyde fractions will increase considerably with increasing denuder temperature (Table 4). At the temperatures 298 K and 323 K, the model returns fractions of around 0.5% and 1.3%, respectively; that is, they are increased by roughly a factor of ~3. These observations are consistent with the physical state of formaldehyde at ambient temperature, namely a gas, which is dissolved in the MS particle-phase. Assuming equilibrium, the gas-to-particle concentration ratio is proportional to the Henry's law constant of formaldehyde in the aerosol phase. Using pure water as a proxy for the smoke aerosol, the temperature dependence of the G/P partitioning of formaldehyde in cigarette MS could be estimated by the temperature dependence of the Henry's law constant of formaldehyde in water (49). Using the data of Allou et al. (50) leads to an estimated ratio of 2.9, in close agreement with the temperature dependence of the initial gas-phase fraction of formaldehyde from the Lipowicz and Piadé model (29).

The importance of water in the G/P partitioning of cigarette smoke formaldehyde was observed by Pang and Lewis (44). They determined carbonyl emissions from a number of cigarettes at normal tobacco moisture and in dried form where much of the tobacco water content had been removed through prolonged oven heating. Pang and Lewis found that the gas-phase formaldehyde emissions increased, but the particulate-phase formaldehyde emissions decreased on moving from normal to dry tobacco cigarettes. The G/P formaldehyde ratio was therefore higher (2 times) with dry tobacco cigarettes. Although smoke water content was not measured in Pang and Lewis's study, in general, smoke water content is made up from two sources, tobacco water and combustion water. By reducing tobacco water content from around 13% to near zero, the water content of the smoke can be anticipated to be reduced markedly in moving from normal tobacco moisture cigarettes to dry tobacco cigarettes. Pang and Lewis's observations are therefore consistent with our explanation of formaldehyde/water interactions influencing G/P ratios of formaldehyde in cigarettes smoke. Pang and Lewis (44) also noted that similar gas/particle partitioning ratios of carbonyls were found among all cigarettes and cigars, which means that the findings of the present study are more generically applicable to cigarettes beyond the reference product studied in this work.

G/P partitioning of nicotine in cigarette MS shows a much stronger temperature dependence than formaldehyde. For instance, the initial gas-phase nicotine fraction estimated by the Lipowicz and Piadé model (29) is around 30 times higher at 323 K than that at 298 K (Table 4). Nicotine is a semi-volatile organic and its denuder behaviour in cigarette MS can be rationalised by the theory of absorptive partitioning developed by Pankow and co-workers (51,52,53). Within this theoretical framework, the strong increase of the gas-phase fraction of nicotine with increasing denuder temperature is explained by the high positive temperature dependence of both nicotine release from salts in the particle [Equation 1] and the vapor pressure of nicotine (25, 54).