Effect of Propylene Glycol on Cigar Mainstream Smoke TNCO Yields

Cigar diameter and draw resistance measurements were performed using a Borgwaldt KC Size Tester S10 and a Draw Resistance Meter A 11 Z (Borgwaldt, Hamburg, Germany), respectively.

Tobacco total alkaloids (reported as nicotine) were determined by continuous flow analysis following CORESTA Recommended Method (CRM) No. 85 (14).

Water content in tobacco was determined by CRM No. 57 (15).

For the determination of cigar moisture content, as oven volatiles, CRM No. 76 was used (16).

Before smoking, cigars were conditioned following CRM No. 46 (17). The analysis of cigar smoke yields is based on CRM No. 65 (18). A Cerulean smoking machine SM410CV (Cerulean, Milton Keynes, UK) was used.

The filter pads were extracted with isopropanol containing as internal standards n-heptadecane for nicotine and ethanol for water. For the experimental cigar analysis, 1,3-butanediol was added as internal standard for PG.

Aliquots of the extracted sample were used to determine nicotine (CRM No. 66 (19)), water (CRM No. 67 (20)) and PG (CRM No. 60 - (21)).

CO was quantified according to CRM No. 68 (22).

PG content analysis in whole cigar tobacco (filler + wrapper) is based on the CRM No. 60 (21).

Three experimental Cigars A, B and C were produced with the same fermented tobacco blend, routinely used for flavored Toscanello^© production for varied PG content targets of 0%, 3.11% and 8.50%, respectively.

Thus, 2 kg tobacco for each type were treated with 400 mL of water containing 0%, 22% and 81.5% of PG, respectively.

In detail, the three different solutions were applied by spraying with a manual nebulizer using a cement mixer under constant stirring to ensure right delivery of PG.

After nebulization, the three tobacco batches were sealed in airtight plastic containers to prevent PG loss and left for two days to ensure homogeneous diffusion of PG through the tobacco. Throughout the process of tobacco preparation, the weight and the relative humidity of tobacco samples were monitored. Cigars were prepared by using MST standard production processes and transferred to a conditioning chamber set at 22 °C and 60% relative humidity.

After conditioning, 6 cigars for each cigar type were used to evaluate water and PG content in whole cigar tobacco, while 30 cigars for each cigar type were submitted to complete smoking analysis, including TPM's (Total particulate matter) PG content analysis.

Statistical analyses were performed by using Minitab 20 Statistical Software (2021) (www.minitab.com).

The differences of the means were evaluated by analysis of variance (One-Way ANOVA) at alpha = 0.05, with Bonferroni correction for multiple testing. If the ANOVA showed a statistically significant result, comparison of the means was performed by Tukey pairwise comparisons procedure (95% confidence).

Several flavored brands are produced and commercialized under the Toscanello^© trade name.

They are machine-made, singly-wrapped cigars with a truncated cone shape and are made entirely with Kentucky dark fire-cured tobacco. They have similar physical characteristics and the same filler and wrapper tobacco but are flavored with different blends.

In Table 1 and Figure 1 the main physical characteristics and a picture of a typical flavored Toscanello^© are reported. Three brands produced with flavoring blends containing low (Cigar M1), medium (Cigar M2) and high (Cigar M3) concentrations of PG were selected for data analysis. The physical characteristics of these brands are reported in Table 2.

Figure 1.

Picture of a typical flavored Toscanello^©.

The mean and 95% confidence interval (CI) of the cigar PG content were 3.11% (95% CI: 2.95; 3.27) for Cigar M1, 5.71% (95% CI: 5.42; 6.01) for Cigar M2 and 8.50% (95% CI: 8.23; 8.77) for Cigar M3, as determined by gas chromatographic analysis of 12 samples for each brand.

Tobacco and tobacco smoke yields data from our quality control laboratory, produced in a period of 4 years, have been evaluated by statistical analysis.

The means of total alkaloids as nicotine, expressed in % on a dry weight basis, obtained from 25, 32 and 51 batches (Cigar M1, M2 and M3, respectively) were M1 2.448% (95% CI: 2.322; 2.574), M2 2.415% (95% CI: 2.290; 2.540) and M3 2.483% (95% CI: 2.410; 2.558). No statistically significant difference was found, as expected given that the brands are made with the same tobacco blend. Afterwards we performed a statistical analysis on the smoke yields data. The analysis was performed on a total of 175, 222 and 324 cigars for M1, M2 and M3, respectively.

Values and plots of means and 95% CI, expressed in milligram per cigar, of CO, nicotine, water and NFDPM are reported in Table 3 and Figure 2, respectively. In Table 3 TPM and puff number are also reported.

Figure 2.

Plots of means and 95% CI, expressed in milligram per cigar, of CO, nicotine, water and NFDPM in mainstream smoke of M1, M2 and M3 cigars.

The statistical analysis showed differences between the three cigars for all the analytes.

We can therefore state, with good confidence, that the presence of increasing quantities of PG results in smaller quantities of nicotine and increasing quantities of CO and water in cigar mainstream smoke. For NFDPM the trend seems quite ambiguous; the three values are statistically different, but they don't show the roughly linear correlation with PG contents shared by the other analytical parameters. Obviously, the analysis conducted on “mg per cigar” basis could be affected by the weight of cigars and possible differences in tobacco content.

Furthermore, it is clear that the greater the presence of PG, other components of the flavoring blend and water, the lower the tobacco content.

To normalize the smoke yields on the basis of the actual tobacco content, we would have to subtract the weight of all substances other than tobacco from the weight of the cigar.

Unfortunately, no PG or other flavoring additives content analysis was routinely performed. Only data on the pre-smoking cigar moisture content, determined by CORESTA Recommended Method No. 76, were available (16).

It must be emphasized that, as stated in CRM 76, the results of moisture content determination by oven drying may be higher than the results of water content analysis when using a specific method. The difference is due to the loss of volatile materials, other than water, among them most of the flavoring blend, including PG.

So, in this kind of study, the normalization for the dry weight obtained by volatile substances determination could be a good compromise to get as close as possible to the actual values of the tobacco content.

By determining the dry weight and normalizing all smoke yields per gram of tobacco, mean and 95% CI of CO, nicotine, water and NFDPM were calculated and are reported in Table 4.

The results from statistical analysis were quite like those obtained without tobacco weight normalization.

The analysis confirmed that all the smoke analytes were different between the three cigars, except NFDPM, for which no statistically significant difference was found between M1 and M3 by Tukey pairwise comparison.

As for statistical retrospective analysis, it must be emphasized that the flavoring blends differ not only in PG contents, but also in some other additives which could play a role in the variation of smoke yields.

With the aim of avoiding any interference by substances other than PG, experimental cigars consisting only of tobacco and water with different concentrations of PG were produced and conditioned. The main physical characteristics of the experimental cigars are reported in Table 5.

Tobacco from 6 cigars for each sample were analyzed for total alkaloids. The percentage of total alkaloid on dry weight basis were 2.46%, 2.43% and 2.45%, respectively for Cigars A, B and C, without statistically significant difference.

An additional 6 cigars for each sample were used to analyze water and PG content by gas-chromatography. Mean values expressed as percent of cigar weight are reported in Table 6.

30 cigars for each sample were selected for smoke yields analysis.

Cigars were smoked by Cerulean SM410CV smoking machine and analyzed.

Values and plots of means and 95% CI of CO, nicotine, water and NFDPM are reported in Table 7 and Figure 3, respectively. In Table 7 TPM and puff number are also reported.

Figure 3.

Plots of means and 95% CI, expressed in mg/cigar, of CO, nicotine, water and NFDPM in mainstream smoke of A, B, and C Cigars.

Statistical analysis showed that the three cigars differ for all the smoke yields but nicotine, for which cigar B showed no difference to Cigar A.

Therefore, if we look at the values of analytes expressed per cigar, we can state that the presence of increasing quantities of PG determine increasing amount of all smoke analytes, except for nicotine.

As for nicotine, higher quantity of PG led to a reduced delivery of nicotine to smoke, while smaller quantity of PG seems to have no effects.

Unlike the retrospective statistical analysis on M1, M2 and M3 cigars, in this case, where only water and PG were added to tobacco, we can make a more accurate estimate of the actual amount of tobacco present in cigars and we can better define the relationship between smoke yields and cigar tobacco amounts.

Subtracting PG and water amounts, as previously determined, from the weights of the 30 smoked cigars, we can calculate the actual amount of tobacco in each cigar and can express smoke yields on a “g of tobacco” basis. Means and 95% CI of smoke analytes expressed in mg/g of tobacco, are reported in Table 8.

Statistical analysis confirmed that the three cigars differ in all smoke analytes except nicotine for which no differences were found between the three cigars.

It seems that the reduced amount of nicotine in Cigar C, shown by analysis on mg/cigar basis, as previously reported (4, 5), is simply due to a reduced presence of tobacco, while the increase in all other smoke yields is actually related to the presence of PG.

It is pretty difficult to define how PG determines the variations in CO, water and NFDPM.

It is very probable that the presence of PG causes some modification of the combustion conditions that influence the way the cigars are smoked, and this leads to an increase of water, CO and NFDPM. Regarding the increase in CO, the possibility that carbonyls, like propanal and acetone, could be produced by thermal degradation of PG and further be degraded to CO should also be considered. The identification and quantification of potential thermal decomposition products of PG are certainly relevant from a regulatory point of view and therefore deserve to be further investigated in future work.

Humectants have the effect of drawing water into tobacco making it more available for vaporization, and this could contribute to increasing the water content in the smoke. But the displacement of tobacco amount by PG, as with nicotine yield, may also play a role.

In fact, although the % of water content per cigar shown in Table 6 was quite the same for the three samples, the water content per gram of actual tobacco was quite different. These values were respectively 91.1 mg, 93.5 mg and 108.6 mg.

Therefore, the water content per gram of tobacco in the Cigar C was higher than that of the other two cigars.

A final aspect we wanted to investigate is the possible contribution of PG to NFDPM yield.

It is known that PG doesn't undergo excessive pyrolysis during smoking and is transferred to cigarette smoke in its pure form.

Baker and Bishop (23) reported that 86.3% of transferred PG (and 6.2% 1,3-propylene glycol) was in its pure form. Purkis, Mueller and Intorp (24) stated that under pyrolysis conditions, propylene glycol was transferred 99.4% intact.

Obviously, PG is partly retained in the filter pad during smoking, contributing to TPM and NFDPM yields.

It is also known that only a small percentage of PG is transferred to smoke during pyrolysis.

In a set of analyses performed on ventilated cigarettes for the PRIORITY ADDITIVES TOBACCO CONSORTIUM it was found that 0.9–0.97% of PG is transferred into the mainstream gas phase (6).

A previous study demonstrated much higher transfer rates of 7.3–8.8% using unventilated cigarettes (24).

In this study, we determined PG content in the TPM of the 30 smoked cigars, as described in the section Materials and Methods.

The mean value of PG in Cigar B is 2.59 mg (95% CI: 2.47; 2.71), representing 5.86% of TPM, while in Cigar C it is 7.60 mg (95% CI: 7.18; 8.02), representing 14.28% of TPM.

We can estimate that 2.1–2.2 % of the PG of the whole cigar tobacco has been delivered intact to TPM.

By subtracting measured PG content from the NFDPM, we obtained the value of NFDPM consisting of all particulate substances except PG (NFDPM-PG).

For statistical analysis we normalized the values of NFDPM-PG per gram of tobacco. In Table 9 the NFDPM/g and NFDPM-PG/g values are reported (obviously, Cigar A didn't contain PG).

While statistical analysis for NFDPM/g showed differences between the three samples, in the case of NFDPM-PG/g no difference was detected between Cigars A and B.

So, in the case of small amounts of PG in the cigar, it seems that the increase of NFDPM is simply due to the delivery of PG into the mainstream smoke, while the presence of higher quantities of PG in tobacco leads to an increase in NFDPM for other reasons as well.

We can deduce that the increase in NFDPM yields is partly due to the mere presence of pure PG in NFDPM and partly due to some mechanism that is complex to elucidate.

It should also be emphasized that the presence of PG might affect moisture and burning properties, perhaps leading to modified NFDPM constituents' levels simply due to the modified combustion conditions.

In our experience, we have often noticed that the presence of higher quantities of PG in tobacco reduces the need for relighting during cigar smoking.