Activated coconut shell charcoal (20–40 mesh, 1100–1200 m²/g surface area) was purchased from Fisher Scientific. Combustible, cellulose acetate-filtered cigarettes were purchased in June 2015 from greater metropolitan Atlanta area retail outlets (Table 1). All cigarettes were stored in freezers at −12 °C (short-term) or −30 °C (long-term).

Products constituted popular cigarette products representative of the marketplace diversity with regard to tar yield, length, and mentholation. Five products were king-size and five were 100s. Two products were menthol cigarettes. Each product was assigned a product code (a–j) used for identification throughout the text (Table 1). The products used in the present study were selected from a group of 50 popular cigarette products with ‘tar’ levels that were quantified previously under the CI smoking regimen (Vu et al., 2015). Tar yields reported previously are not identical to those reported in the present study due to slight differences to the smoking regimen employed (Section 2.4.). Product characteristics and tar yields are available in Supplementary Table 1 and Fig. 1, respectively.

Filter segments of the experimental cigarettes were modified to contain a central cavity containing charcoal granules between two plugs of cellulose acetate filter in a PSP configuration (Fig. 2). For each modified cigarette, the two cellulose acetate (CA) plug portions of the filter were equal in length. First, the cigarette filter was cut in half perpendicular to the cigarette length. The portion of CA adjacent to the tobacco rod was separated from the portion of CA at the mouth end of the cigarette (CA2) by a charcoal-containing cavity. Pre-weighed charcoal loadings were placed between CA1 and CA2, and held in place with non-porous tipping paper and adhesive tape positioned to block ventilation holes on the un-modified cigarette tipping paper. The charcoal-containing cavity was the minimum length required to contain the charcoal granules. The modified king-size cigarettes included 100 mg, 200 mg, 300 mg, or 400 mg charcoal, while the modified 100s cigarettes included 100 mg, 200 mg, or 300 mg charcoal. Due to the extent of carbonyl reduction observed for king-size cigarettes, it was deemed unnecessary to include the 400 mg charcoal condition in 100s cigarette testing. Unmodified cigarettes of each of the 10 products served as control conditions (0 mg). Control cigarette filters were not modified, but all filter ventilation holes were covered by nonporous tape. Experiments were conducted in two sequential phases wherein king-size cigarettes were constructed and analyzed first, followed by construction and analysis of 100s cigarettes. Although there have been reports of release of charcoal granules during smoking (Pauly et al., 1997), observation of the mouth end of experimental cigarettes after machine smoking did not suggest charcoal release in the present study. All experimental and control cigarettes were conditioned for at least 48 h at 22 °C and 60% relative humidity prior to machine smoking.

All cigarettes were smoked on a linear smoking machine under a fixed puff count CI smoking regimen to represent and make comparisons within scenarios with the highest potential constituent yields. For carbonyl quantification, three cigarettes were machine smoked per charcoal loading. For TNCO analysis, three cigarettes were machine smoked for each of three replicates, for a total of nine cigarettes per condition. Cigarettes were smoked to a prescribed number of puffs, which was one puff less than the standard 3 mm plus overwrap procedure indicated by the CI smoking regimen (HC, 1999). This adaptation was made to ensure that all cigarettes for a particular product have identical smoke volumes, despite modifications to the filter segment. Puff counts used for each product are included in Tables 2 and 3.

Carbonyl analysis was based on a recently published method on the derivatization and quantification of 7 carbonyls in cigarette MSS (Ding et al., 2016). Prior to smoking, Cambridge filter pads (CFP) were pre-treated with derivatization reagent. Cigarettes were smoked through the treated pads according to the fixed puff count CI smoking regimen on a linear smoking machine. MSS carbonyls were derivatized and collected on the pre-treated CFPs. Derivatized carbonyls were extracted in the solvent and an aliquot of the sample was diluted and injected into an ultra-high pressure liquid chromatography coupled with a triple quadruple mass spectrometer (UPLC-MS/MS).

Cigarette rod length and filter plug were measured manually using a Vernier caliper. Charcoal amount for each cigarette was manually weighed using an analytical balance before it was assembled into the filter segment. Cigarette draw resistance was measured on the C² instrumentation (Cerulean; Milton Keyes, UK).

The standard procedures previously described for TNCO determinations were used with few modifications. Briefly, the gas and particulate phases of MSS from 3 cigarettes per sample were collected in vapor phase collection bags and on CFPs, respectively. The percentage by volume of CO (%CO) was determined from vapor phase collection bags using a built-in COA205 non-dispersive IR analyzer. The total particulate matter (TPM) was determined gravimetrically by calculating the weight difference of the CFP before and after smoking divided by the number of cigarettes smoked per pad. The TPM was then extracted with 20 mL of extraction solution (isopropanol containing approximately 0.1 mg/mL anethole and 1 mg/mL methanol ISTDs) by gently shaking at 160 rpm for 30 min. A blank, conditioned CFP was extracted concurrently with smoke samples for background water subtraction. The extract was then analyzed for nicotine and water by gas chromatography-flame ionization detection (GC-FID) and GC-thermal conductivity detection (GC-TCD), respectively. Calibration curves were constructed with 10 different analyte concentrations (ranging from 0.004 to 1.0 mg/mL for nicotine and 0–5.0 mg/mL for water) plotted against the area ratios of analyte-to-internal standard. A linear regression analysis (1/X weighed) of the calibration curve provided the slopes and intercepts from which the nicotine and water concentrations of unknown sample extracts could be calculated. A multiplier (i.e., number of cigarettes smoked per CFP divided by the extraction solution volume used) was then applied to calculate nicotine and water amounts in mg/cigarette. The determined water content of the blank CFP was then subtracted from all samples. Finally, tar content was derived by subtraction of the determined water and nicotine contents from TPM.

Two-tailed student’s t-tests were used to determine statistical significance of carbonyl reduction at each charcoal loading. The average yield of each carbonyl across all cigarettes of each length class (king-size or 100s) were compared for each charcoal loading versus the carbonyl yield of king-size or 100s cigarettes without charcoal (Table 4). Tests used the nominal level of significance of α = 0.05, and results were considered significant when p ≤ 0.05.

MSS carbonyl yields were quantified for all cigarettes (Fig. 3). For every product, significant reductions of all carbonyls were observed with increasing charcoal mass. For control products, acetaldehyde levels were highest, followed by acetone, and levels of the remaining five carbonyls were less. Although significant reductions in formaldehyde yields were observed (median 32% reduction at 100 mg charcoal), the effect of charcoal filtration on formaldehyde yields was least substantial among yields of other carbonyls (median 80.5% reduction at 100 mg charcoal). The percent reductions of carbonyl yields compared to the control products are tabulated for king-size (Table 2) and 100s products (Table 3).

The most substantial decreases in carbonyl yields occurred between controls and the 100 mg charcoal condition, with p ≤ 0.01 for all carbonyls except formaldehyde, where p = 0.13 for king-size products and p = 0.03 for 100s products. Notably, among king-size and 100s products, crotonaldehyde decreased by an average of 91% and 87%, respectively, between the control and the 100 mg charcoal condition. For king-size products, the absolute yield of crotonaldehyde between the control and the 100 mg charcoal condition decreased by greater than seven-fold from 42.9 μg/cig to 5.6 μg/cig. Similarly, for 100s products, the absolute crotonaldehyde yield between the control and the 100 mg charcoal condition decreased nine-fold from 47.6 μg/cig to 5.2 μg/cig.

Acrolein yields for the 100 mg charcoal-filtered cigarettes decreased between 81% and 79% on average among king-size and 100s products, respectively. For king-size products with 100 mg charcoal, the absolute acrolein yield decreased by greater than fourfold, from 130.8 μg/cig to 28.8 μg/cig. Similarly, for 100s products with 100 mg charcoal, the absolute acrolein yield decreased by five-fold, from 124.2 μg/cig to 24.5 μg/cig, on average.

With 200 mg charcoal, the cumulative percent reduction of all carbonyls increased across all king-size products (p < 0.01 for all carbonyls except formaldehyde, where p = 0.02) and 100s products (p ≤ 0.0015). For example, crotonaldehyde yields of king-size and 100s products with 200 mg charcoal decreased by 96% and 94%, respectively, from control 100s cigarettes. At the 200 mg charcoal level, several data points were below the limit of quantification (Fig. 3).

With 300 mg charcoal, carbonyl levels either decreased further or remained relatively consistent to those observed at the 200 mg level, since reductions of several carbonyls began to approach 100% with higher charcoal loadings. For example, with 300 mg charcoal, the cumulative percent reduction of crotonaldehyde was 98%, on average, across all king-sized products and 94%, on average, across all 100s products, elevated only slightly from or similar to the 96% and 94% cumulative percent reductions observed for the 200 mg charcoal condition. For king-size products with 300 mg charcoal, all carbonyls had significantly lower yields than controls (p ≤ 0.001 for all carbonyls except formaldehyde, where p = 0.005). For 100s products with 300 mg charcoal, all carbonyls had significantly lower yields than controls (p < 0.001).

With 400 mg charcoal, carbonyl levels of king-size cigarettes approached limits of quantification (p < 0.001 for all carbonyls except acetaldehyde, where p = 0.0075). However, many of the carbonyl levels observed for the 400 mg charcoal condition were below the LOD for the detection assay, indicating that the additional charcoal reduced the carbonyl levels further, despite those levels being as much as 98% less than those of the controls.

TPM and TNCO yields were quantified for king-size (Fig. 4A) and 100s (Fig. 4B) cigarettes. There were no statistically significant reductions in mean TPM or TNCO yields between the control and charcoal-filtered king-size cigarettes at any of the prescribed charcoal amounts. For 100s cigarettes, there were no significant reductions in TPM and TNCO for cigarettes containing 100 mg charcoal versus control cigarettes. However, with 200 mg and 300 mg charcoal, there was a statistically significant reduction in TPM compared to controls for 100s cigarettes (p = 0.03 and p = 0.004, respectively). Additionally, at the highest charcoal level for 100s cigarettes (300 mg), there was a statistically significant decrease in tar (p = 0.007) and nicotine (p = 0.009), relative to controls. Fig. 5 illustrates the relative percent reductions in TNCO (dark shaded bars) and carbonyl (light shaded bars) levels in charcoal-filtered cigarettes versus control cigarettes in king-size (Fig. 5A) and 100s (Fig. 5B) products. Across all products filtered with 100 mg charcoal, the median reductions of tar and nicotine were 16% and 8%, respectively. Comparatively, with 200 mg charcoal, the median reductions of tar and nicotine were 20% and 19%, respectively.

Cigarette draw resistance is a measure of pressure differential across the length of a cigarette, is correlated with smoking behavior, and influences puffing topography and consumer acceptance of cigarettes. In this study, cigarette draw resistance was quantified for king-size (Fig. 6A) and 100s (Fig. 6B) control and charcoal-filtered cigarettes. Draw resistance is tabulated in Tables 2 and 3 for king-size and 100s cigarettes, respectively. Among king-size cigarettes, there were no significant increases in draw resistance with 100 mg or 200 mg charcoal loadings. Draw resistance increased significantly compared to controls at higher charcoal loadings of 300 mg and 400 mg (p = 0.02 and p = 0.006, respectively). There were no significant increases to draw resistance for 100s cigarettes at any charcoal loading. P-values for all draw resistance comparisons versus control cigarettes for king-size and 100s cigarettes are included in Table 4.

The effect of charcoal filtration on carbonyl yields was similar across all products analyzed, and carbonyl yields decreased with increasing charcoal mass (Fig. 3). For all products, there were substantial reductions in all carbonyls, even at the lowest charcoal loading (100 mg). The space portion of the modified cigarettes required approximately 3 mm to accommodate 100 mg charcoal, suggesting that substantial carbonyl reduction may be achieved with minor changes to cigarette design and appearance.

Reductions in carbonyls increased with increasing charcoal load, and plateaued at approximately 99% reduction at the higher charcoal loadings for all products and for all carbonyls except for formaldehyde. As demonstrated previously (Branton et al., 2011b; Dittrich et al., 2014), formaldehyde was reduced to a lesser extent than other carbonyls. Branton et al. propose that its relatively high vapor pressure at room temperature may reduce formaldehyde absorption by charcoal (Branton et al., 2011a, 2011b). In addition, the difficulty of MSS formaldehyde quantification has been noted by several, which merits further investigation as it relates to charcoal filtration of mainstream smoke (Parrish and Harward, 2000; Thomas and Koller, 2001). We are also aware of the possibility of alternate reactions of formaldehyde and acetaldehyde which may lead to the formation of reaction products (methanediol and lactonitrile, respectively) with decreased volatility (Dube and Green, 1982; Johnson et al., 1966; Rickert et al., 2007; Socrates, 1969). Accordingly, the measurements of formaldehyde and acetaldehyde may slightly underrepresent the absolute quantities of these constituents in the smoke. Nevertheless, the differences in formaldehyde and acetaldehyde yields in charcoal-filtered versus control conditions indicates substantial reduction of all constituents that were examined. Although the present study demonstrates substantial reductions in acetaldehyde in charcoal-filtered cigarettes, alternate filtration materials may be required to obtain similar reductions of formaldehyde, and altering the configuration and composition of charcoal granules may improve removal of a wider range of volatile aldehydes from MSS (Branton et al., 2009, 2011a). Additionally, future research may necessitate the use of alternate quantification methods to optimize formaldehyde detection in MSS of charcoal filtered cigarettes. Accordingly, the observed trend of limited formaldehyde removal efficiency by charcoal may be due to both the chemical and physical nature of formaldehyde and its interaction with charcoal, meriting further investigation.

A loading of 180 mg charcoal was the highest charcoal loading found in published studies that analyzed smoke composition of charcoal-filtered cigarettes under the CI smoking regimen (Polzin et al., 2008; Rees et al., 2007). Polzin et al. quantified yields of several volatile smoke constituents and TNCO from four commercially available test market charcoal-filtered cigarette prototypes, one of which used 180 mg charcoal in a PSP filter format. With 180 mg charcoal loading, the MSS contained 574.4 μg/cig acetaldehyde and 64.3 μg/cig acrolein. By comparison, the MSS of the non-charcoal containing control cigarettes contained 1221 μg/cig acetaldehyde and 235.3 μg/cig acrolein. As such, 180 mg charcoal removed 53% of the acetaldehyde and 73% of the acrolein from the cigarette MSS.

Despite experimental differences between Polzin’s work and this study, reductions in acetaldehyde and acrolein of the previous work align with the results observed here. To illustrate this point, the reductions of acetaldehyde and acrolein in the MSS of cigarettes modified to contain 100 mg and 200 mg charcoal in the present study were compared to the reductions in acetaldehyde and acrolein observed by Polzin et al., with 180 mg charcoal. Here, 100 mg charcoal removed 50% of acetaldehyde (1542.1 μg–764.0 μg) and 80% of acrolein (127.5 μg–26.7 μg) from the across all products. With a higher loading of 200 mg charcoal, 78% of the acetaldehyde (1542.1 μg–339.3 μg) and 92% of acrolein (127.5 μg–10.2 μg) was removed from the MSS of all products. These results are consistent with previously reported studies and indicate that a substantial portion (>50%) of carbonyl yields can be removed with between 100 mg and 200 mg charcoal (Polzin et al., 2008; Rees et al., 2007).

We assessed TNCO yields of control and charcoal-filtered cigarettes, as they are important properties of cigarette smoke (1988; Gunby, 1988). As expected, CO was not significantly influenced by charcoal filtration at any loading, since charcoal has not been associated with removal of CO in the gaseous state. While tar and nicotine yields decreased to some extent with increasing charcoal in all products, the removal of tar and nicotine by charcoal was modest. Due to the well-documented non-selective nature of charcoal filtration, modest removal of tar and nicotine was not surprising (Rahman et al., 2006). Substantial reductions in carbonyls without proportional TNCO reductions suggest that, especially for lower charcoal loading conditions, substantial decreases in carbonyls may be possible without impacting nicotine deliveries.

Although the modified cigarettes were intended to provide a proof-of-concept for charcoal filtration of carbonyls in MSS, we analyzed draw resistance to determine whether charcoal inclusion would cause technical design limitations for modified cigarettes. We observed statistically significant increases to draw resistance at higher charcoal loadings (300 mg and 400 mg) only. A recognized limitation of charcoal filtration is the potential for draw resistance increase to render the cigarette unpleasing to the consumer (Raker et al., 1996). The draw resistance of 50 popular currently marketed cigarettes (Vu et al., 2015) was between 83 mmH₂O and 194 mmH₂O (unpublished data). In this case, several charcoal-containing cigarettes had draw resistance values within the range of draw resistance that is found in currently marketed cigarettes that do not have filter modifications. For example, products a, d, and j modified with 100 mg charcoal had draw resistance values of 139 mmH₂O, 130 mmH₂O, and 133 mmH₂O, each of which is within the range of draw resistance that is observed in currently marketed cigarettes. As such, several cigarettes modified to contain 100 mg and, in some cases, 200 mg and 300 mg charcoal, exhibited substantial carbonyl reduction while incurring only moderate increases to draw resistance compared to currently marketed cigarettes.

Reduced-exposure cigarettes likely require integration of complementary design modification technologies to achieve reduced harm to health. Several cigarette design adaptations, including increased smoke dilution, novel tobacco substitutes, alternate filter adsorbents, and alternate filter tow plasticizers, have been investigated for their combined effect on constituent reduction (Dittrich et al., 2014; Liu et al., 2011; McAdam et al., 2011, 2012). In addition, filter configuration and design can be optimized to achieve target levels of constituent reduction. Generally, to achieve optimal reduction of volatile smoke constituents, the interaction between the smoke and the surface of the charcoal particles or alternate adsorbent should be maximized (Branton et al., 2009). Approaches toward this goal include optimization of the filter design format to ensure that the smoke passes through a sufficiently dense bed of charcoal, avoiding empty pockets of space within the cavity region (Pauly et al., 1997). Including alternate, high surface area nano-materials, such as carbon nanotubes, may yield similar carbonyl reductions without adversely influencing draw resistance or other critical design parameters. Additionally, more detailed information regarding the influence of charcoal filtration on smoke flavor would aid the design of reduced-exposure cigarettes. To this point, prototype charcoal-filtered cigarettes with alternate flavor systems, such as a flavored cotton thread woven into the filter and crushable menthol beads, have been used to enhance acceptability or overcome flavor changes caused by charcoal filtration (Gordon et al., 2011; Rees et al., 2007; Strasser et al., 2013).

Although machine yields provide insight to constituent exposure to human smokers, without toxicological and human biomarker studies, potential exposure reduction cannot be used alone to determine health outcomes of reduced exposure cigarettes of this study. Therefore, although outside of the scope of the present work, additional research on product shelf life and stability, as well as consumer perception and behavioral studies, toxicity studies, and human biomarker levels, will help to predict health outcomes that may be expected when consumers use reduced-exposure products. Ultimately, a combination of technologies or adaptations could be formulated to construct a reduced-exposure cigarette without sacrificing consumer acceptability. The proof-of-concept presented here may inform the design of reduced exposure cigarettes intended for consumer use.

This work suggests the potential for substantial reduction of select harmful smoke constituents without concurrent broad sacrifices to other cigarette design parameters. Combined, these data indicate the potential for an achievable cigarette design modification that may reduce exposure to some carbonyls. At higher charcoal loadings, carbonyl yields of modified cigarettes have improved our understanding of the utility of charcoal filtration in combustible cigarettes and the limits associated with such modifications, as several carbonyls were nearly completely removed from MSS at high loadings. The knowledge gained here may also be used to inform technologies that may further reduce yields of other HPHCs in cigarette smoke.