Determination of Hydrogen Cyanide in Mainstream Cigarette Smoke, CORESTA Collaborative Study and Recommended Method

The laboratories that conducted the testing are listed in Table 1. To ensure anonymity of the results, each laboratory was given a unique laboratory code for data exchange and reporting.

Participating laboratories were asked to follow a supplied study protocol which included a draft of the CORESTA Recommended Method (CRM) for the determination of HCN. Samples analyzed for the study, listed in Table 2 were selected to achieve a wide range of “tar” values (1 mg–16 mg). Participants were instructed to target five replicates in a specified randomized run order and to report puff count, total particulate matter (TPM), total HCN, and optionally, pad HCN and impinger HCN.

Following the recommended method, the cigarettes were smoked according to ISO 4387 (17). Typically, if using a linear smoking machine, 4 cigarettes were smoked per replicate for ISO 3308 (ISO non-intense) and 2 cigarettes per replicate for ISO 20778 (ISO intense) smoking regime. If using a rotary smoking machine, the cigarettes smoked were 10 and 20 cigarettes per replicate, respectively. In any case, a sufficient number of cigarettes for detection without risking particulate breakthrough for the trapping system were required. Cigarettes were smoked using standard smoking machines fitted with “Pad+Imp” (trapping system 1) or “Pad+Pad” (trapping system 2).

For trapping system 1, the cigarette holder with one glass fiber filter pad (Cambridge filter pad; CFP) was connected in tandem to a fritted stem impinger (Figure 1). The CFP and fritted stem impinger were used to trap particulate- and gas-phase HCN of smoke, respectively. 30 mL of 0.1 M sodium hydroxide solution is suitable for ISO non-intense regime and 90 mL of 0.1 M sodium hydroxide solution for ISO intense regime. After smoking, the CFP was removed, folded into a quarter (with the condensate inside) and used to wipe the inside of the holder. The CFP was placed in 50 mL of 0.1 M sodium hydroxide solution (44-mm pad) or 100 mL of 0.1 M sodium hydroxide solution (92-mm pad). The flask was shaken at 200 rpm on a wrist-action shaker for 30 min. The extract solution was filtered into the sample vessel by a 0.45-µm membrane filter. The impinger solution was quantitatively transferred into a volumetric flask (e.g., 100-mL, 200-mL). The impinger was triple rinsed with measured 0.1 M NaOH(aq). The flask was filled to volume with 0.1 M NaOH(aq). The mixed solution of impinger and rinsing solution was transferred to the sample vessel.

Figure 1.

Example of a suitable trapping system 1.

In trapping system 2, the treated CFP was positioned in line with a blank CFP in the cigarette holder connected directly to the machine piston (Figure 2). When using a linear smoking machine, a 44-mm CFP was placed into a 60-mm culture disk or a 100-mL beaker containing 2.0 mL of a solution of 1 M NaOH (in 50:50 EtOH:water). After being fully saturated, it was put into a fume hood for about 10 min to evaporate ethanol and then transferred into a controlled environment for 1–3 h at 22 °C ± 1 °C and 60% ± 2% of relative humidity. When using a rotary smoking machine, a 92-mm CFP was placed into a 100-mm culture disk or a 500-mL beaker containing 8.0 mL of a solution of 1 M NaOH (in 50:50 EtOH:water). The procedure of evaporation and conditioning of the 92-mm CFP were the same as for the 44-mm CFP. The cigarette holder was subsequently loaded with the CFPs with rough sides of both CFPs toward the incoming smokestream.

Figure 2.

Example of trapping system 2.

Post smoking, the two CFPs were disassembled, folded together into a quarter (with the condensate inside) and used to wipe the inside of the holder. The two CFPs were placed into 100 mL of 0.1 M NaOH(aq) solution (44-mm pad) or 250 mL of 0.1 M NaOH(aq) solution (92-mm pad). The flask was shaken at 200 rpm on a wrist shaker for 30 min. The extract solution was filtered into the sample vessel through a 0.45-µm membrane filter.

HCN in CFP extracts and impingers was determined using CFA based on the detection of the coloring system as specified by the manufacturer. For example, a system formed by the reaction of cyanides with chromogenic reagents, including pyridine/pyrazolone and 1,3-dimethyl barbituric acid. A single-channel spectrometer (e.g., 600 nm) was employed to detect the colored complex. The manifold of a continuous flow analyzer and the typical output of HCN determination are shown in Figure 3 and 4, respectively.

Figure 3.

The manifold of the continuous flow analyzer for hydrogen cyanide determination.

Figure 4.

Typical graph of hydrogen cyanide determination by CFA.

Eighteen participating laboratories provided 22 data sets across both method subtypes (“Pad+Imp” and “Pad+Pad”) using either linear (18 sets) or rotary (4 sets) smoking machines. Data consistency was verified using numerical outlier detection techniques of IUPAC (18). Grubbs test was used to evaluate between-laboratories consistency and Cochran’s test were used to evaluate within-laboratory consistency. Because the number of data sets (labs) using “Pad+Imp” trapping system was no more than five, it was unnecessary to use IUPAC method to perform the outlier detection for these labs. The results of outlier detection are listed in Table 3. Under the ISO non-intense smoking regime, only one or two labs corresponding to each sample were excluded, while under the ISO intense smoking regime, the excluded labs corresponding to each sample ranged from 1–4. Statistical analysis was carried out following ISO 5725-2 to generate repeatability (r) and reproducibility (R) data on a combined data set without regard for sample collection, or detection type or model.

The study design allowed for evaluation of the robustness of the method across the linear range for analytes of interest and to confirm the method’s effectiveness with a variety of products. Thus, the study samples chosen represented a wide variety of ISO 3308 “tar” levels, tobacco blends, and product construction parameters including varied filter ventilation level and inclusion of the charcoal filter. ISO non-intense and intense smoking regimes were used to verify applicability with the recommended method. Indeed while “tar” determination was not in the scope of the study, the range of experimentally determined total particulate matter (TPM) values was quite broad at approximately 1mg/cig to 50 mg/cig.

Table 4 and Table 5 summarize the mean HCN yields and standard deviation (Stdev) of the seven samples under ISO non-intense and ISO intense smoking regimes, respectively. The mean HCN yield decreased with decreasing ISO 3308 “tar” levels under ISO non-intense regime. The relative variabilities (%RSD, percent relative standard deviation) from trapping system 2 were smaller than those from trapping system 1 under both smoking regimes. The paired t-test was used to compare mean smoke yields of HCN achieved by the two trapping systems. No significant differences in results were noted between the data from the two trapping systems under either ISO non-intense (p = 0.555) or ISO intense (p = 0.304) smoking regimes. In most cases, results from the two trapping systems were consistent with each other, which suggests either the “Pad+Pad” or “Pad+Imp” method subtype can be adopted in the determination of HCN in mainstream cigarette smoke.

The r and R values under ISO non-intense and ISO intense regimes are displayed in Table 6 and Table 7, respectively. After the removal of outliers, r and R values were calculated from data sets of various trapping systems, i.e., “Pad+Pad”, “Pad+Imp”, “Overall” (combined data for both trapping systems). Under the ISO non-intense smoking regime, both r and R values of the “Pad+Pad” trapping system were lower than those of the “Pad+Imp” trapping system. The maximum r(%) of “Overall”, “Pad+Pad”, and “Pad+Imp” trapping systems were 20.0%, 19.7% and 24.4%, respectively. The maximum R(%) of “Overall”, “Pad+Pad”, and “Pad+Imp” trapping systems were 32.3%, 30.6% and 68.4%, respectively. The lowest ISO 3308 “tar” yielding products had the most variable data.

Under the ISO intense smoking regime, the maximum r(%) of “Overall”, “Pad+Pad”, and “Pad+Imp” trapping systems were 11.3%, 10.8% and 22.4%, respectively. The maximum R(%) of “Overall”, “Pad+Pad”, and “Pad+Imp” trapping systems were 25.6%, 26.7% and 37.8%, respectively. Except for the r values of Sample 2 and 4, both r and R values of the “Pad+Pad” trapping system were also lower than those of the “Pad+Imp” trapping system.

Based on ISO 13528 (19), z-scores are designed to indicate whether lab results are within the normal range of other labs. It is expected that most z-scores should be in the range of −2 to 2. A value of 2 ≤ |z| < 3 should be interpreted as a warning signal, and a dataset with a value of 3 ≤ |z| should be considered an “action signal” to examine laboratory performance. Figure 5 shows the z-scores for each lab calculated from the “Overall” data sets. Z-score results exhibit only two data sets, both of which came from “Pad+Imp” trapping systems, that were determined to be outliers.

Figure 5.

Z-scores for each lab under ISO non-intense and ISO intense smoking regimes.