Article Text

Download PDFPDF

Existing technologies to reduce specific toxicant emissions in cigarette smoke
  1. R J O’Connor,
  2. P J Hurley
  1. Department of Health Behavior, Roswell Park Cancer Institute, Buffalo, New York, USA
  1. Richard J O’Connor, Department of Health Behavior, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, NY 14263, USA; Richard.Oconnor{at}


Background: The World Health Organization Tobacco Product Regulation (TobReg) study group has proposed emissions level performance standards for nine toxicants (NNN, NNK, acetaldehyde, acrolein, 1,3-butadiene, CO, BaP, benzene and formaldehyde, all expressed as micrograms per milligram nicotine as measured under the Canadian intensive method) in cigarette smoke for parties to the FCTC in conjunction with regular monitoring of emissions of nine other toxicants of interest, nicotine and nicotine-free dry particulate matter (NFDPM, or “tar”).

Methods: We examined the published literature and publicly available tobacco industry documents to determine the extent to which existing available technologies can be applied to reduce the emissions of the specified toxicants in cigarette smoke.

Results: Agricultural practices (for example, fertilisers, curing), plant characteristics (for example, protein content, nicotine content), tobacco blending (for example, American blend vs Virginia blend) and cigarette design (for example, additives, filters, paper) issues all have roles in the generation and reduction of specific smoke toxicants. The tobacco industry has explored a number of technologies, including selective filtration, changes to curing practices and rod additives to reduce specific toxicants.

Conclusions: Technologies exist to reduce the toxicants identified by TobReg. The extent to which the industry is able to simultaneously reduce toxicants, however, is unknown.

View Full Text

Statistics from

The public health community in Western countries assumed for decades that nominal changes in tar and nicotine yields as measured on smoking machines translated into exposure and disease reductions in humans. However, the history of “light” cigarettes shows this not to be the case.1 2 Manufacturers developed strategies to increase product elasticity, facilitating the ability of smokers to extract desirable levels of nicotine regardless of the cigarette’s machine rating.14

Under Articles 9 and 10 of the World Health Organization’s Framework Convention on Tobacco Control (FCTC), parties will be developing systems to regulate the contents, design and emissions of products and to require reporting on the same from manufacturers. Several countries currently have limits on tar and nicotine emissions (for example, European Union, Brazil, South Africa, Egypt) and others require regular reporting of tar and nicotine emissions and ingredients (for example, Canada, United States, Hong Kong). The Conference of Parties has begun deliberations as to the best way to recommend how parties deal with tobacco product regulation. In recognition of the importance of this issue, WHO in 2003 convened a formal Study Group on Tobacco Product Regulation (TobReg) to examine the scientific basis for regulation. After several meetings, TobReg has issued a report5 proposing emissions level performance standards for nine mainstream smoke toxicants, with the precautionary aim of reducing known toxicants having variability in the existing market. These constituents, listed in table 1, were selected because they are known animal and/or human toxicants, though reductions in their emissions are not yet proved to yield population health benefits. The group also recommended that emissions measurements be made using the Canadian Intensive smoking regimen, which takes larger (55 ml), more frequent (30 seconds) puffs than the traditional Federal Trade Commission (FTC)/International Organization for Standardization (ISO) regimen (35 ml, 60 seconds) and also employs full blocking of filter vents.

Table 1 Toxicants recommended for mandated lowering by TobReg5

An important aspect of shaping the future of tobacco product regulation is to consider how the tobacco industry might respond to future restrictions on product contents and emissions. It is expected that any lowering of specific constituents will require changes to the construction of cigarettes, whether in the tobacco blend, additives, filter or paper. This paper examines the published literature and publicly available tobacco industry documents for evidence of existing technologies that manufacturers could use to comply with proposed limits. Our goal here is not a comprehensive review, but rather to illustrate methods that manufacturers could use, based on their previous research. This should also not be taken as a recommendation or endorsement of specific technologies, but simply as an overview of selected methods examined in the past.


Data were obtained from the published literature using systematic searches of the US National Library of Medicine’s PubMed database, the Beitrage zur Tabakforschung International archive ( and the US Department of Agriculture literature database. In addition, we conducted a search of tobacco industry documents on the two major interface sites (; Finally, we searched for patents relevant to the reduction of the toxicants of interest using Google Patents. Key search terms included the constituent names, “filter”, “blend”, “precursor”, “reduce”, “emission” and similar words and phrases.

It must be noted that, for historical reasons, the majority of data cited in this report are machine-smoking data obtained the using the FTC/ISO method, rather than the Canadian Intense method recommended by TobReg. Nevertheless, the FTC/ISO data may provide some insight into general approaches. Where possible, attempts have been made to standardise reductions per mg tar or nicotine or per unit volume—most data are presented as per-stick emissions, but it is important to note that per-stick emissions differences can be misleading unless identical products are being compared.


Table 2 provides a list of examples of technologies applied to the reduction of each of the identified constituents. Each is discussed in more detail in the paragraphs below.

Table 2 Summary of example technologies explored to reduce emissions of toxicants targeted for mandated lowering


The tobacco-specific nitrosamines NNN (N'-nitrosonornicotine) and NNK (4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone) are potent carcinogens and the wide disparity in nitrosamine levels observed among brands internationally (for example, Counts et al,6 Ashley et al,7 Wu et al,8 Fischer et al,9 Gray et al10) suggests differences in blending, curing methods and/or manufacturing processes are the major contributors to levels of nitrosamines in tobacco and tobacco smoke.11 12 NNN tends to be the most predominant tobacco-specific nitrosamine (TSNA) in American blend cigarettes while NNK is most predominant in Canadian flue-cured tobacco cigarettes.6 In general, the TSNA content of mainstream smoke can be predicted by the TSNA content of the tobacco rod, whether a single grade or blended. TSNAs are consistently higher in burley than in bright tobacco, but flue curing of bright tobacco produces nearly three times the TSNA levels compared to air-curing. The use of nitrate fertilisers can increase the concentration of both alkaloids and nitrosamines in tobacco.13

For flue-cured tobaccos, TSNA probably result from the reaction of combustion gases used in the heating process of curing with tobacco alkaloids, with microbial action playing a lesser role. For air-cured tobaccos, such as burley, curing and post-curing treatment, genetics, alkaloid concentration and storage all impact TSNA levels.14 The biological mechanism for TSNA formation in air-cured tobacco relates to the breakdown of plant cell membranes caused by moisture loss, making cell contents available to nitrate-producing micro-organisms—that is, microbial action generates nitrate as a byproduct, which, when converted to nitrite, becomes available to react with alkaloids to produce TSNA. Strains of tobacco can differ in their rates of moisture loss and also in their propensity to convert nicotine to nornicotine (precursor to NNN). Leaves cured in well ventilated curing structures (for example, plastic sheds), generally contain lower TSNAs than leaves cured in barns. However, TSNA may increase post-cure if tobaccos are stored under humid conditions or in bales.15

Approaches to reducing TSNA levels in smoke fall into three broad categories: agricultural practices, curing and tobacco blending. Given existing evidence, it appears that reducing the use of nitrate fertilisers would lead to a generalised reduction in TSNAs. Changing the heating source away from propane for flue-curing appears to have reduced TSNAs in recent years for this type of tobacco.11 12 It is likely that shifting away from direct exposure of leaves to combustion exhaust during flue-curing has resulted in a lowering of TSNAs in flue-cured Canadian tobacco. For air-cured tobaccos, TSNAs could be lowered by curing in well-ventilated structures at cooler temperature and low relative humidity, provided that the nornicotine content of the variety is low, nitrogen fertilisation during growing was limited and the time from curing to threshing is as brief as possible.

Based on the available data, the simplest way for manufacturers to reduce TSNA emissions in the short term would be switching to blends consisting primarily of bright, flue-cured tobacco, as in Canada and Australia. These blends have TSNA levels that are now up to 10-fold lower than American blended cigarettes, which contain air-cured burley and other tobaccos. However, actions that could further reduce TSNA in markets where flue-cured tobacco predominates are currently unclear. As new curing and other technologies reduce preformed TSNA (that is, in the tobacco rod), formation via pyrosynthesis will become a more important consideration. Finally, the use of specialised filters, additives, paper type or porosity or other physical parameters do not appear to substantially affect TSNAs in smoke. Irwin, in a 1990 internal review for BAT, reported that filtration efficiencies for NNN and NNK varied closely with total particulate matter (TPM) and nicotine, suggesting no selectivity.16

Carbonyl compounds—acetaldehyde, acrolein, formaldehyde

Acetaldehyde, acrolein and formaldehyde share similar structures and chemical properties. Therefore, they should share common methods for reduction in smoke and are grouped together accordingly.


Acetaldehyde arises from the combustion of carbohydrates, particularly cellulose and added sugars (see fig 1).17 18 Carbohydrates (cellulose, pectins, starches, casings and simple sugars) can represent as much as 30% of the mass of a blended cigarette.19 20 Casings are sugar solutions often added to tobaccos such as burley that tend to lose sugar during curing.21 22 Sugars generally add flavours and act as humectants in blended tobacco.2024 Other contributors to acetaldehyde concentration identified in the past include tobacco pH, leaf thickness, percentage cellulose and stalk position.2527 Cellulosic materials may account for up to 50% of acetaldehyde in smoke.27

Figure 1 Relation between sugar content and aldehyde yield. (Data from Phillpotts et al.24)

Zilkey et al28 examined cigarettes prepared from tobacco types differing in sugar levels (that is, no added sugars). They reported that sugar levels accounted for over 50% of the variance in smoke acetaldehyde levels. Phillpotts of BAT reported that for 40 commercial UK brands, sugar content and moisture were unrelated to acetaldehyde yield, though acetaldehyde was related to TPM yield.24 Similar associations were reported for brands from continental Europe. Re-analysis of the pooled data suggests that, if analysis is limited to filtered brands only, sugar content accounts for 23% of variability in aldehyde levels (β = 0.48, p<0.001) and that sugar content is related to overall tar level (β = 0.37, p<0.003). Published industry reports have generally normalised acetaldehyde yields to tar or TPM—these studies report no correlation between tobacco sugar content and smoke yields of acetaldehyde (reviewed by Seeman et al18). When we adjust the Phillpotts data for tar, we also find no relation. However, if one treats the problem multivariately, one sees a different pattern. If TPM is forced into the model first, it accounts for 23% of variance in aldehyde yield (β = 0.48, p<0.001). This makes sense given TPM for filter cigarettes would be a surrogate for design features such as ventilation as well as mass of tobacco (which was not reported). If one then adds sugar content to the model, it is a significant predictor (β = 0.35, p<0.004) and accounts for an additional 11% of variance in aldehydes and does not render TPM non-significant (β = 0.35, p<0.004) by virtue of shared variance. So, normalising for tar may obscure a sugar-aldehyde association.

A number of options have been explored within the tobacco industry to reduce acetaldehyde emissions, including limiting the addition of exogenous sugars to tobacco and using blends of tobaccos containing fewer sugars.29 In mainstream smoke deliveries, charcoal-filtered cigarettes have been shown to significantly reduce many volatile components, including acetaldehyde, with the amount of volatile compounds removed varying directly with the amount of charcoal present.29 30 Rodgman cites a RJ Reynolds combination of charcoal and bauxites as removing up to 30% of acetaldehyde from smoke under traditional FTC conditions.31 Polzin and coworkers examined commercially available cigarettes with varying levels of charcoal (0 mg, 45 mg, 120 mg, 180 mg) but similar to FTC tar delivery (5–6 mg) and other design parameters.32 Cigarettes were tested under Canadian Intense conditions and emissions were then normalised to nicotine (consistent with the TobReg proposal). The highest carbon content filter showed the greatest reduction in acetaldehyde relative to the non-charcoal equivalent, with substantial amounts of charcoal (180 mg) required to reduce acetaldehyde significantly (approximately 45%). Laugeson and Fowles report no difference in acetaldehyde under Canadian Intensive conditions for two Mild Seven versions containing charcoal versus a similar comparison brand without carbon.33 34 In fact, when normalised to nicotine, Mild Seven Light actually showed the highest level of acetaldehyde.

Philip Morris USA patented a filter that is claimed to selectively reduce aldehydes by up to 78% from mainstream smoke “comprising a granular carrier containing concentrated hydrogen peroxide, water and a hydrophilic stabiliser for said hydrogen peroxide.”35 B&W tested Duolite, a resin containing primary, secondary and tertiary amine groups, under the hypothesis that acetaldehyde would be removed from smoke via addition to primary and secondary amines.36 37 B&W examined the resin’s properties, filtration effectiveness and particle release during manufacture, packaging and smoking (including inhalation and pyrolysis.36 38 The Fact “low-gas” cigarette for which it was developed was briefly on the US market in the 1970s. In 2002, B&W introduced Advance Lights, featuring a “Trionic” filter comprising an ion exchange resin, charcoal and cellulose acetate. The resin was purported to interact with “semi-volatile constituents of tobacco smoke, particularly … aldehydes (including formaldehyde) and others like hydrogen cyanide”. The carbon section was claimed to”adsorb the aldehydes from the smoke, as well as trapping other constituents like benzene and acrolein”. The overall product claim was for overall reduced toxins relative to other light cigarette brands, prompting calls for FDA regulation of the product.39


Acrolein is known to be a decomposition product of glycerol (glycerin) when it is heated above 280°C. Glycerol is commonly used as a humectant in tobacco, applied at concentrations of 1% to 5% of total weight and so may account for a portion of the acrolein found in smoke. The production of acrolein from humectants was known in the 1930s, as evidenced from a letter sent by Philip Morris to physicians along with a pamphlet describing the effects of acrolein and its derivation from glycerine.40 Memos dating from the same period show that American Tobacco was measuring acrolein in smoke41 and examined paper versus tobacco cigarettes and salt additives to paper as sources. A 1960 Philip Morris report42 noted that addition of 3% by weight of glycerol to bright tobacco led to a 10% increase in acrolein in smoke in specially prepared research cigarettes (that is, no other additives used, matched for tar delivery). The report also notes that glycerin could only account for about 8.5% of the acrolein content of smoke. Gager and colleagues43 summarised this work, adding that the values observed were lower than those reported for commercial cigarettes, likely “because of the particular tobaccos used and the presence of other additives (flavours, etc) …” Blake and coworkers at Philip Morris Europe44 investigated the relation of glycerin and propylene glycol as precursors to aldehydes in experiments using commercial blend cigarettes and differing levels of glycerin, propylene glycol or mixtures thereof as humectants. They calculated that 1% of available glycerin was converted to acrolein.

In a more recent study by Philip Morris USA, Carmines and Gaworski examined effects on smoke chemistry of cigarettes containing three levels of glycerol (5%, 10% and 15%).45 Higher glycerol content (10–15%) was associated with a statistically significant increase (9%) in smoke acrolein, but this effect was not seen at the 5% application level of glycerol. When these data are adjusted for nicotine delivery, one sees a 6.5% increase in acrolein at the 5% application level, a 33% increase at the 10% application level and a 44% increase at the 15% application level, all compared to the control cigarette of identical blend and construction.

In general, design changes that would reduce acetaldehyde would tend to reduce acrolein. Data from the study by Polzin et al show that acrolein can be reduced as much as 69% under Canadian Intense conditions (normalised to nicotine) when a high-carbon filter (120 mg–180 mg charcoal) is used.32 Laugeson and Fowles report a difference in acrolein of approximately 15% for Mild Seven varieties (which contain far less charcoal than the varieties tested by Polzin) when compared to a non-charcoal filtered brand at Canadian Intense conditions—this held true whether or not emissions were normalised to nicotine.33 34

In the 1970s, Lorillard looked to reduce acrolein by adding reactive compounds to the filter and/or tobacco rod, such as hydrazines, high molecular weight amines, ammonium phosphate and ammonium carbonate.46 By 1976, some additives that selectively reduced acrolein had been identified, but a memo indicates none was patentable47 and the work appears to have been abandoned.


Formaldehyde originates from saccharide materials, such as sugars and cellulose.17 4852 Baker concluded that “all sugars added to tobacco and cellulose, increase the yield of formaldehyde in mainstream cigarette smoke”. Studies reviewed by Baker observed increases of up to 60% at the highest sugar levels tested, with similar effects observed under more intensive smoking regimens.52

Reducing or eliminating exogenous sugars from tobacco processing (for example, casings) should reduce formaldehyde smoke yields. Carbon filters can effectively reduce formaldehyde in smoke; Laugeson and Fowles reported an approximately 33% reduction in formaldehyde under Intensive smoking for Mild Seven cigarettes compared to a non-charcoal brand (normalised to nicotine).34 For a 180 mg charcoal test-market brand (Marlboro Ultrasmooth), they report reductions of 75% in formaldehyde compared to Marlboro regular when not normalised to nicotine, but this dropped to 51% when adjusted for nicotine.35 36 A memo and patent from Philip Morris53 54 describes the use of gamma-aminopropylsilane-silica gels (APS-SG) to selectively filter formaldehyde from cigarette smoke in model systems.

Ammonium compounds and amino acids have been reported to inhibit the generation of formaldehyde from sugars,52 via reactions that form hexamethylenetetramine (HMTA).55 A 1990 RJ Reynolds memo55 describes results of a study examining effects of including ammonia sources (urea and glycine) within the tobacco rod on MS yields of various carbonyl compounds. The authors report that formaldehyde reductions ranging from 70–98% were found, depending on the level of urea or glycine added.


Benzene in smoke appears to form from the thermal decomposition of more complex organic molecules such as amino acids, waxes, complex sugars and cinnamic acid,56 57 and levels appear to be directly related to the amount of tobacco burned in the cigarette.

Polzin and colleagues observed a filter containing 120 mg of charcoal or more can reduce benzene emissions by nearly 85%32 even under Intensive conditions and normalised to nicotine. This observation in machine generated smoke is consistent with human exposure data reported by Scherer and colleagues,58 although the mass market charcoal cigarettes studied would probably contain far less than 120 mg charcoal. Smoking carbon-filter cigarettes was associated with statistically significant reductions in urinary benzene exposure biomarkers (for example, S-phenyl mercapturic acid). Searches of the tobacco documents and patents failed to find other methods specifically tested for the reduction of benzene in mainstream smoke.


Polycyclic aromatic hydrocarbons (PAHs) can be generated from most organic compounds given sufficient heat and quantity.59 60 Over 500 different PAHs have been completely or partially identified in tobacco smoke.61 Benzo[a]pyrene (BaP) is always a component of the mixture and can serve as a surrogate marker for other carcinogenic PAHs.12 Some PAH can be measured on cured tobacco leaves, though this is probably accumulation from environmental sources (including curing) rather than a component of the tobacco. Rodgman and Perfetti provide a chronology of papers related to four PAHs in smoke, including BaP.61 The general consensus reached over time is that large molecules such as phytosterols, long-chain aliphatic hydrocarbons, terpenoids, amino acids, fatty acids and waxes, collectively referred to as “extractable components,” account for approximately 60% of PAH levels.56 6267 Tobacco types appear to differ up to tw fold in their BaP emissions, with flue-cured and Oriental producing more than burley or Maryland (see for example, Wynder and Hoffmann68).

Since PAHs were identified in mainstream smoke in the 1950s, methods to reduce their levels have been investigated. Rodgman and colleagues at RJ Reynolds extracted saturated aliphatic hydrocarbons, phytosterols and terpenoids (such as solanesol) from tobacco to decrease PAHs in mainstream smoke.61 However, a side effect of this extraction process was an increase in the per-unit-weight content of carbohydrates (including cellulose), which are precursors of aldehydes. By the 1960s, it was shown that higher levels of nitrates in the tobacco led to lower PAH levels in smoke, probably through the ability of nitrate to interfere with the free-radical processes that generated PAH. Wynder and Hoffman68 postulated that observed drops in BaP yields over time might be due to the introduction of reconstituted tobacco, particularly recon made from stems (which are relatively high in nitrates); Halter and Ito found similar results.69 However, since nitrates contribute to TSNA formation, this may have in essence traded one class of carcinogens for another. Rodgman summarised the various changes to cigarettes at various stages of production explored by RJ Reynolds to reduce PAH yields in smoke, shown in table 3.59

Table 3 Methods explored to reduce PAH yields

The “XA” cigarette examined by Liggett employed treatment of tobacco with nitrate salts containing a palladium catalyst to reduce the biological activity of smoke. In the 1977 patent issued to Liggett workers for the palladium additive,70 it is mentioned that the additive reduced PAHs in smoke. A 1992 “Attorney Work Product”71 gives a good summary of testimony culled from the trial. Liggett never marketed a cigarette using the palladium technology, which was resurrected in the Omni cigarette sold by Vector Tobacco in the early 2000s. Vector claimed 2–42% lower PAH emissions (depending on measurement method) compared to conventional cigarettes. However, human studies showed that those who switched to Omni did not show reduced PAH exposure relative to their usual brand (as measured by urinary 1-hydroxypyrene).72 73

A recent development in the reduction of PAH are “free-radical traps” to inhibit the formation of PAHs during pyrolosis or selective filtration of PAHs. Fillagent patented a porphyrin additive for filters that would trap PAHs and this appears to have been introduced into the new Fact cigarette.74 75 However, few data are available on this product. Another possibility for selective reduction was a DNA-based solution presented by Lodovici and colleagues.76 They report that the solution, which added approximately 10 mg DNA to a cellulose acetate filter (extracted from salmon sperm) could bind with PAH, including BaP, as smoke passed through the filter.77 Lodovici et al report 40% decreases in BaP in treated compared to control cigarettes under ISO conditions.76 However, it is unclear whether this level of reduction is reproducible given that most PAHs do not react with DNA until reduced to epoxides (which is not known to occur in smoke).


A review performed for Philip Morris suggests that butadiene may arise in smoke from pyrolisis of waxes, sterols, terpenoids and other large organic molecules.78 Studies of cellulose and Cytrel cigarettes showed that cellulosic materials are unlikely to be major sources of butadiene.79 80 Ferguson postulates that the total lipid or wax content of tobacco can be used as an index of butadiene precursors; similarly, the extractables (discussed earlier under BaP) may be a reasonable indicator. But, extraction may prove difficult as a way to reduce yields given 1,3-butadiene may be formed by a number of different pathways.

Limited data were located on selective reduction of butadiene. Activated charcoal has been suggested as a possible method. RJ Reynolds in the 1990s explored a “carbon scrubber” filter technology—a 7-inch piece of wood pulp impregnated with 40 μm particles of carbon gathered to provide channels of 0.2–0.7 mm.81 82 The technology was supposed to reduce gas phase constituents including acrolein, butadiene and formaldehyde. One internal R&D report83 notes that butadiene per mg wet particulate matter was lower in a modified Camel Light cigarettes using the carbon scrubber in conjunction with higher filter ventilation (a 24% reduction). Adding potassium carbonate to the blend increased the percent reduction to approximately 48%. However, another analysis using “straight-grade” cigarettes showed no significant difference in butadiene.84 This research on carbon scrubber filters went as far as human laboratory studies,8486 and a proposal for a 90-day inhalation study in rats.87 It appears to have evolved into a disposable filter attachment (that is, an aftermarket accessory) rather than a filter used at manufacture.88 It is possible, of course, that the reported reductions could be the result primarily of filter ventilation rather than the carbon additive.

Carbon monoxide

Carbon monoxide (CO) is a primary product of the incomplete combustion of carbon-containing materials.89 As such, there are no specific chemical precursors to the gas in tobacco. Baker estimates that 30% of CO comes from thermal decomposition of tobacco components (for example, starch, cellulose, sugars), 36% by combustion of tobacco and 23% by carbonaceous reduction of carbon dioxide (CO2).90 The main factors that influence variability in CO yield relate to burning temperature and availability of oxygen (that is, completeness of combustion), as this will drive the ratio of CO2 to CO.

CO has been a target for emissions reduction for some time. Unlike other gas phase constituents, however, most studies show that activated charcoal is ineffective in significantly reducing CO in mainstream smoke.91 Polzin et al showed that even 120 mg or more of charcoal, while lowering other gas phase volatiles (see earlier sections) had no significant impact on CO.32 Irwin notes that CO “is present in smoke at substantially greater concentrations than other smoke components … Hence substantial quantities of substances which combine chemically with it will be required to effect filtration”.16 Tolman92 summarised a number of possible techniques for removing CO via selective filtration:

  • Chemisorption in a microporous subtrate (molecular sieve)

  • Physical adsorption

  • Chemical complex formation

  • Absorption in a liquid in which CO is soluble

  • Oxidation to CO2

  • Catalytic oxidation in which oxygen is consumed

  • Disproportionation to carbon and CO2

  • Reduction to carbon.

Tolman9294 appears to have led a substantial work programme at BAT focusing on selective filtration of CO in the early 1970s. In summarising this work, Irwin notes that depositing manganese dioxide in the pores of activated carbon appeared to reduce CO emissions, but that the exothermic CO to CO2 conversion catalysed by MnO2 heated the charcoal to the point that it glowed red.16 Irwin concluded that there was “little evidence of effectiveness” of complexing agents, concluding that a more promising research direction was molecular sieves.16

BF cigarettes, featuring the “Bio-Filter”, which claims that haemoglobin embedded in the filter sequesters carbon monoxide, were introduced in Greece in 1997,95 96 based on a design described by Deliconstantinos et al.95 Valvinidis and Haralambous reported less CO removal than was achieved in the original paper (30–35% vs 90%).96 Researchers at Lorillard attempted to reproduce the method, but were unsuccessful,97 speculating that “failure of the proteins to alter the delivery of the measured components may have resulted from the incorporation of a smaller than adequate amount of each protein. A larger amount of each protein could not be incorporated using the Deliconstantinos method since the protein solutions used were at the saturation limit.” Tolman had calculated that 10–20 g of haemoglobin on filter would be required to remove 15–20 mg of CO from smoke.93 94 Researchers at Philip Morris have also explored haemoglobin filters.98 Patents for haemoglobin or other absorbent additives to the filters are available99102 but report relatively small changes to CO yields.

Researchers at RJ Reynolds examined palladium salts to reduce CO in smoke, even exploring this for the Premier device.103 Lewis summarised their investigations of palladium/copper catalysts added to the filter and showed that they could reduce CO in smoke by as much as 20%.104 However, concerns over cost, availability of sufficient materials and toxicology appear to have caused them to abandon the work.

Filter ventilation was introduced on a large scale in part because it was the easiest way to reduce CO levels in smoke.105 However, given the use of the Canadian Intense smoking regime, which blocks vents, as the basis for regulation, ventilation is not a feasible option to reduce yields in this context. The use of higher porosity cigarette paper, which would make more oxygen available and promote the diffusion of gaseous components out of the burning rod, has been shown to reduce CO output of cigarettes.90


A review of tobacco industry literature shows that tobacco companies have at various times investigated technologies for reducing emissions of all of the constituents identified by TobReg. Changes to agricultural practices, tobacco curing, blending, filtration and cigarette construction could be combined to bring about the targeted reductions over time. However, it is difficult to say with certainty what combination of approaches would be employed to simultaneously reduce the targeted emissions. Indeed, tobacco companies are currently preparing for shifts in product regulation in the United States (under the Food and Drug Administration) and other countries (under FCTC), including proposals for emissions disclosures and product performance standards. As noted in a recent special issue of the industry trade journal Tobacco Reporter “…there will be broader regulation of the tobacco industry in the future and, in part, the industry will no doubt welcome this development”.106

Because the proposed limits are based on data for the distribution of toxicant levels for a sample of brands available on the market, there is a subset of existing brands that would already comply with the proposed limits. Studying the characteristics of these brands can be instructive. When one examines data from the paper by Counts et al6 on emissions of Philip Morris international brands (a dataset used by TobReg to define the recommended levels of 125% of the median value for most toxicants and the median value for NNN and NNK), one sees some interesting patterns of results for brands that are below the recommended levels for all toxicants. Nine brands from 48 reported by Counts and colleagues’ list would be below the recommended levels for all nine toxicants.6 Interestingly, none features a charcoal filter. When one examines how these nine differ from the remaining brands on the physical characteristics and rod contents reported in the paper, one sees only two statistically significant differences (all p values from Mann-Whitney U test). The brands with levels below the recommended values have lower nitrate content (6.98 mg/g vs 8.76 mg/g, p<0.009) and lower NNK (485.4 ng/g vs 851.4 ng/g, p<0.004) content in the blend. No differences are seen for filter ventilation, paper porosity, tobacco weight, puff count (a surrogate for burn rate), blend nicotine content or blend ammonia content. If one ignores NNN and NNK (assuming that blend changes would allow brands to meet the limits), then 27 brands have yields below the recommended levels. These brands differ from the others on only a few characteristics. They had higher blend ammonia (2.03 mg/g vs 1.40 mg/g, p<0.014) and higher puff counts (9.8 vs 8.6, p<0.015), with borderline effects for blend NNN (2309.4 ng/g vs 1785.4 ng/g, p<0.05) and tobacco weight (0.698 g vs 0.655 g, p<0.07). Carbon filtered brands were not more likely to be below the recommended levels (66.7% for carbon filters vs 53.8% for others, p = 0.71 by Fisher’s exact test). The existing brands that appear to meet the proposed limits differ from those that would not, largely, on the contents of the tobacco rod (for example, nitrate, ammonia), rather than cigarette construction parameters such as ventilation or paper porosity. This is probably in part the result of the adjustment of yields to nicotine, which would remove the impact of ventilation and other design “tricks” that reduce smoke concentration per cigarette. It is also important to note that these cigarettes were designed under different regulatory conditions (that is, tar, nicotine, carbon monoxide reporting on a per-cigarette basis), so it is not inconceivable that new regulatory restrictions could encourage certain shifts in cigarette composition or construction over others.

Unintended consequences?

Like all population-level policy interventions, there are trade-offs involved and potential risks in implementation. Of particular concern are potential unintended consequences associated with pursuing emissions limits. A lesson of FTC/ISO tar-based reporting should be that, like any profit-driven enterprise, the cigarette industry will find ways to comply with regulations while changing the fundamental nature of their products as little as possible. So, monitoring the contents and design characteristics of products over time, pursuant to Article 10 of FCTC, will be crucial to knowing how manufacturers are complying. One possibility for complying with limits based on ratios of toxicants to nicotine is to manipulate the denominator—that is, manufacturers can increase nicotine yield (for example, by adding nicotine to the rod or filter) rather than finding routes to reduce individual toxicants. TobReg recommends monitoring nicotine yields and tar-to-nicotine ratios in the market over time to look for this type of manipulation. Beyond this, governments might also consider regulating nicotine content and/or emissions in conjunction with TobReg’s targeted toxicants.

Cigarette smoke is a complex aerosol containing upwards of 4000 different chemicals of varying concentrations and toxicities. While constituents fall into broader chemical classes (carbonyls, phenolics, PAHs, TSNAs), changes to tobacco agriculture, curing, blending and/or cigarette construction intended to influence a specific set of constituents are almost guaranteed to have consequences for other classes of compounds. And, in some cases, these ancillary effects may be in an unfavourable direction—decreasing one set of toxins causes another set to increase. Vigilant monitoring of emissions, including toxicants of health concern not presently targeted for limits (such as hydrogen cyanide, 4-aminobiphenyl and catechol), is essential to demonstrate that these negative side effects are minimised.

One of the lessons of the public health disaster of the light cigarette is that the appearance of safety is a powerful influence over smokers.107113 The negative population health consequences of mistakenly accepting the purported benefits of lower tar and nicotine products have been significant.114116 Yet, it appears that tobacco industry analysts view the low-tar segment as a robust area for future growth.117 118 In particular, The shift towards lower tar cigarettes can be abetted by well meaning governments publishing ISO/FTC tar and nicotine yields on packs, which merely draws attention to meaningless distinctions among brands.

Carbon particle emissions

One way for the cigarette industry to comply with emissions limits on volatile organic compounds (aldehydes, benzene, butadiene) could be to implement wider use of charcoal filters, particularly filters featuring higher levels of carbon than is typical on most currently marketed brands. Recent tobacco trade publications have noted a trend towards greater use of charcoal and predict more widespread use of charcoal and other “granular additives” to reduce vapour phase components.119 However, as Pauly and colleagues have pointed out, this has a potential downside.120 121 Charcoal filters can emit numerous small inhalable carbon particles during puffing, which may become embedded in lung tissue. Philip Morris has been aware of carbon particle fallout since the 1960s and developed methods to measure it.122124 Agyei-Aye et al of Brown and Williamson performed examinations of charcoal filters under sham smoking conditions (similar to Canadian Intensive conditions, except for a 20-second puff interval) and found a number of particles given off during sham smoking, of which tobacco flake and charcoal were the most abundant.125 The authors do not report the weight or number of charcoal pieces in each brand, so it is difficult to determine the extent of transfer. Of the 102 particles measured for the highest particle yield Advance prototype, 44 (43%) were 200 μm2 or less in surface area. As noted in a 1977 internal memo at RJ Reynolds, “Charcoal-filtered cigarettes with substantially reduced aldehydes (as well as other gas-phase components) have generally not fared well in the [US] marketplace …”31 However, charcoal filters have had much greater market shares in countries like Japan and South Korea. Early epidemiology pointing to lower cancer rates in Japanese smokers was sometimes attributed to charcoal filters; however, more recently analyses have been inconclusive (for example, Muscat et al126 and Stellman et al127). Data on the health impacts of regular charcoal filter use are difficult to find. While it is unclear what the health implications of a shift towards widespread or universal use of charcoal filters would be, regulators may want to examine carbon fall-out issues as part of routine product characteristic reporting and monitoring.


The findings of this literature review suggest that technologies exist now to reduce emissions levels of the toxins targeted for regulation. However, careful routine monitoring of smoke emissions in local markets is essential to detecting potential unintended consequences of the regulations. Some tobacco control advocates fear that proposing emissions limits risks repeating the low-tar disaster. As consumers become more aware of health dangers associated with smoking cigarettes, they become targets for cigarettes marketed with explicit or implicit claims of reduced health risks.117 118 It is a somewhat perverse consequence that educating smokers about the health risks of smoking represents a marketing opportunity for the cigarette industry. TobReg’s proposal seeks to address some of the shortcomings of previous attempts at reducing emissions by focusing on specific compounds and adjusting for nicotine yields. A way to address the potential misinterpretations is to issue strong proscriptions against the use of testing data in marketing, claims or on packs. However, this still leaves other avenues, such as colour and imagery, open to the industry to turn regulation to its advantage, which will necessitate strong and consistent implementation of FCTC provisions on product labelling, advertising and marketing. And, while great care has gone into setting limits that would be difficult to circumvent by easy design manipulations, it should not be assumed that the cigarette industry hasn’t already considered its options—available literature suggests about a 30 year head start.

What this paper adds

  • The WHO Study Group on Tobacco Product Regulation (TobReg) has proposed emissions performance standards for nine mainstream smoke toxicants selected because of their known animal and/or human toxicity. However, the extent to which existing available technologies can be applied to reduce the emissions of the specified toxicants in cigarette smoke is less well known.

  • Examination of tobacco industry documents and published literature reveals that technologies exist now to reduce emissions levels of the toxins targeted for regulation. Changes to agricultural practices, tobacco curing, blending, filtration and cigarette construction could be combined to bring about the targeted reductions over time. However, it is difficult to say what combination of approaches would be employed to simultaneously reduce the targeted emissions.


Thanks to Mirjana Djordjevic, David Burns and David Hammond for comments on earlier drafts.


View Abstract


  • Funding: This paper is adapted from a working paper commissioned by the WHO Study Group on Tobacco Product Regulation (TobReg) for their 4th meeting, 23–25 July Stanford, CA, USA, and WHO supported RJO’C’s travel to the meeting. The work was also supported via the Roswell Park Transdisciplinary Tobacco Use Research Center (TTURC; 1 P50 CA111236).

  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.