Analysis of crotonaldehyde puff-by-puff release in mainstream cigarette smoke under various smoking regimens by high-performance liquid chromatography with the modified QuEChERS method

To study puff-by-puff release characteristics of crotonaldehyde in mainstream cigarette smoke under diverse intensive smoking regimens, we designed an RM20H smoking machine with a puff-by-puff smoke collection unit to automatically trap crotonaldehyde in the mainstream cigarette smoke. Using this process, we trapped, puff-by-puff, crotonaldehyde in mainstream smoke generated by different smoking regimens and quantitatively analysed the levels of crotonaldehyde using high-performance liquid chromatography with a modified QuEChERS sample pretreatment method. On the basis of the crotonaldehyde in each puff, we determined crotonaldehyde's puff-by-puff release characteristics. The results showed that crotonaldehyde's puff-by-puff release remained nearly constant for the International Organization for Standardization mode while increased polynomial trend was seen (n ≥ 6) under the Massachusetts and Health Canada smoking regimens. The equation fit for various regimens was good (R2 > 0.9192). Release characteristics by puff were classified into four categories: (1) first, second and third puffs; (2) fourth and fifth puffs; (3) sixth puff; and (4) seventh and eighth puffs.

Analysis of crotonaldehyde puff-by-puff release in mainstream cigarette smoke under various smoking regimens by high-performance liquid chromatography with the modified QuEChERS method To study puff-by-puff release characteristics of crotonaldehyde in mainstream cigarette smoke under diverse intensive smoking regimens, we designed an RM20H smoking machine with a puff-by-puff smoke collection unit to automatically trap crotonaldehyde in the mainstream cigarette smoke. Using this process, we trapped, puff-by-puff, crotonaldehyde in mainstream smoke generated by different smoking regimens and quantitatively analysed the levels of crotonaldehyde using high-performance liquid chromatography with a modified QuEChERS sample pretreatment method. On the basis of the crotonaldehyde in each puff, we determined crotonaldehyde's puff-by-puff release characteristics. The results showed that crotonaldehyde's puff-by-puff release remained nearly constant for the International Organization for Standardization mode while increased polynomial trend was seen (n (R 2 > 0.9192). Release characteristics by puff were classified into four categories: (1) first, second and third puffs; (2) fourth and fifth puffs; (3) sixth puff; and (4) seventh and eighth puffs.

Introduction
Cigarette smoking machines are used to simulate consumer smoking behaviour, which is important for the study of the chemical composition of mainstream cigarette smoke. Many government regulatory bodies have developed relevant policies and regulations on various parameters of smoking machines according to the rules set by the International Organization for Standardization (ISO) Technical Committee (specifically for tobacco, under the subcommittee of ISO/TC 126) [1]. The ISO standard regimen for machine smoking, like other standard machine regimens, cannot fully represent average human smoking behaviour [2]. As debate on the relevance of the ISO smoking parameters continues in the literature, it is important to study the effect of smoking regimens on chemical release because it affects the combustion of cigarettes [3]. Therefore, study of the effects of different smoking regimens (including the Massachusetts and Health Canada intensive smoking regimens) on puff-by-puff chemical composition in cigarette mainstream smoke has an increasingly important role in the accurate analysis of smoke emission [4]. Currently, in China, much of the research on smoke tar, nicotine, water, ammonia, carbon monoxide, the main compounds of phenols, tobacco-specific nitrosamines, hydrogen cyanide and nitrogen oxides emissions is being conducted under different cigarette smoking regimens [3,[5][6][7][8][9][10][11][12]. However, few reports have described crotonaldehyde's release and the factors that may influence its levels in mainstream smoke. For example, a study by Xie et al. [13]. on cigarette fire temperature distribution described the seven harmful components in mainstream smoke. Li and coworkers [14] studied the release of six chemical constituents by puff in ISO mode using a transformation smoking machine. Chen [15] studied the relationship between important parameters and the release of crotonaldehyde in mainstream smoke by changing key process parameters in the method.
Many methods used to analyse the release of crotonaldehyde in mainstream smoke have been reported, some of which have used high-performance liquid chromatography (HPLC) for quantitative analysis of a whole cigarette [16,17]. Puff-by-puff release analysis of crotonaldehyde in mainstream smoke has not been reported. In particular, when carrying out an extraction process from a Cambridge filter, filter fines are usually suspended in the filtrate, leading to filtration difficulties and severe matrix effects. Therefore, it is necessary to develop a simple, effective and reliable method for quantitative analysis of the puff-by-puff release of crotonaldehyde.
Our study presents a simple HPLC with modified QuEChERS sample pretreatment method for the effective quantification of the puff-by-puff release of crotonaldehyde in mainstream cigarette smoke, under different smoking regimens. Additionally, we used several statistical methods to provide a reference for reducing crotonaldehyde levels in mainstream cigarette smoke.

Specification of cigarette samples
We coded four types of domestic commercial cigarettes as A, B, C and D. Cigarettes B and C were highgrade tobacco with a high rate of total sugar and reducing sugar; cigarettes A and D were low-grade tobacco with a low rate of total sugar and reducing sugar. All four types of cigarettes were 84 mm in length and had a circumference of 24.3 mm; other parameters are given in table 1.

Parameters of ISO, Massachusetts and Health Canada smoking regimens
We used three different smoking modes to study the release of crotonaldehyde. Differences in the puffing parameters among the ISO, Massachusetts and Health Canada smoking regimens included puffing volume, puffing frequency and puffing rate. They all had the same puffing duration of 2.0 s as listed in table 2.

Sample preparation
We followed the requirements of GB/T 19609-2004 [18] for connecting to the RM20H puff-by-puff unit. Twenty cigarettes were included for the traps of total particulate matter from cigarette smoke under the ISO model while 10 cigarettes were included for the Massachusetts and Health Canada smoking modes. We measured each cigarette sample twice and trapped the total particulate matter on Cambridge filters. After the cigarette-smoking using the RM20H puff-by-puff unit, two blank puffs are required. Then the trapper was taken out and left to stand for 3 min. We added DNPH to the filter to react with all carbonyl compounds. A glass fibre filter was used to trap the total particulate matter puffby-puff in the mainstream smoke. After the process was completed, we placed the filter into a 100 ml Erlenmeyer flask and added 50 ml of pyridylacetonitrile. The solution was ultrasonicated for 10 min at 30°C. Then, we immediately vibrated the solution for l min after adding 3 g of sodium chloride and 12 g of anhydrous MgSO 4 . We centrifuged the obtained solution for 5 min and separated 1 ml supernatant. After centrifugation, we simultaneously added 100 mg of anhydrous MgSO 4 , 10 mg GCB, 10 mg C 18 , and 40 mg iron oxide (Fe 3 O 4 ) magnetic nanoparticles [19] into the supernatant. With a 30 s vortex, we collected the supernatants using an applied magnetic field. Treated analytes appeared to be clear and transparent, without any filter impurities; 2 ml supernatant was transferred to a chromatographic flask and we detected the concentration of crotonaldehyde using the HPLC method.
We identified the carbonyl compounds using HPLC. We obtained eight points of carbonyl compounds in the peak area. Using these peak areas as the ordinate, we set concentrations of carbonyl compounds as the abscissa. We set eight kinds of carbonyl compounds by a calibration curve. For the linear regression calibration data, we determined that for each compound R 2 should be not less than 0.990. After determination of the extracted sample, we calculated the eight concentrations of the sample carbonyl compounds according to the area of the fingerprint spectrum of these compounds extracted from the smoke sample of every cigarette.

Data analysis
CHEMPATTERN software (Chemmind Technologies Co., Beijing, China) was used to analyse crotonaldehyde release data in mainstream smoke on a puff-by-puff basis. We used descriptive statistics, scatter trends and multiple correlation analysis to evaluate release trends on a puff-by-puff basis.

Recovery of modified QuEChERS
The validity of the modified QuEChERS method was estimated by means of recovery experiments and all experiments were conducted in duplicate. The average recovery of the target compound ranged from 91.5% to 110.7%, and the relative standard deviation was not more than 5.5%. The recoveries and relative standard deviations were in line with crotonaldehyde analysis requirements.

Chromatographic behaviour
We obtained test results of crotonaldehyde release on a puff-by-puff basis from the cigarette samples and analysed the results under different intense smoking regimens. The retention time of crotonaldehyde was 28.88 min. Standard and sample chromatograms are shown in figure 1 and figure 2, respectively. Furthermore, the resolution, theory plate and symmetry factor of peaks are shown in table 3.       where q L is the minimum detectable concentration, k is a constant related to confidence, s b1 is the standard deviation value for the blank, and s is the sensitivity at low concentrations, which is the slope of the standard curve in the low concentration range (table 4). Setting k at 3, we obtained the slope of the standard curve as s = 0.936. Therefore, the detection limit q L = 3 × 0.0142/0.936 = 0.0455 µg ml −1 . According to the test's dilution and concentration conditions, we obtained the detection limit (per puff) of this method as q L ' = 0.0455 × 50/20 = 0.1138 µg/cigarette. The limit of quantification q L = 0.1138 × 10/3 = 0.379 µg/cigarette was far lower than the crotonaldehyde content of a minimum 1.410 µg analysed per puff in a cigarette. The relative standard deviation of analysed = s.d./mean × 100% = 0.0142/0.1905 × 100% = 7.45%. Therefore, the method can be used to analyse levels of crotonaldehyde in ordinary cigarettes on a puff-by-puff basis.

Descriptive statistics of crotonaldehyde per puff in mainstream smoke under ISO regime
Descriptive statistics of the samples are listed in   In the ISO mode, crotonaldehyde release by puff demonstrates a polynomial trend increase (n ≥ 6), the R 2 = 0.9682, and the proposed combined degree of curve fitting was general. In Massachusetts mode, crotonaldehyde release by puff demonstrates a polynomial trend and the release gradually decreased after the ninth puff, the R 2 = 0.9192 and the degree of curve fitting was ordinary. In the Health Canada mode, crotonaldehyde release by puff also showed a polynomial trend, the release gradually decreased after the ninth puff, the R 2 = 0.9754 and the degree of curve fitting was relatively better.

R-clustering analysis of crotonaldehyde release by puff for different intensive regimens
We used the R-clustering analysis method to study the release characteristics of crotonaldehyde by puff for different intensive regimens (links between cluster groups, using Euclidean distance).

Conclusion
We used an RM20H smoking machine with an automated puff-by-puff trapping unit to collect crotonaldehyde from total particulate matter on a Cambridge filter for three different smoking regimens. We conducted a quantitative analysis using an HPLC external standard technique with a modified QuEChERS sample pretreatment method. We observed the treated analytes to be clear and transparent, without any filter impurities. The retention time of the crotonaldehyde was 28.907 min. The limit of detection (0.1138 µg/cigarette), the limit of quantification (0.379 µg/cigarette) and the RSD (7.45%) all met the detection needs for puff-by-puff release analysis. In ISO mode, the release of crotonaldehyde for the same sample remained almost the same during the puff-by-puff smoking process, but the release varied among the different samples within a range of 1.41-1.72 µg per cigarette. The largest release by puff was from sample A (1.673 µg per cigarette). In the Massachusetts and Health Canada modes, the crotonaldehyde release demonstrated a polynomial trend, and this release gradually decreased after the ninth puff (R 2 ≥ 0.919). The R-cluster analysis showed that the first eight puff features could be divided into four categories according to the different intensive regimens: (1) first, second and third puffs; (2) fourth and fifth puffs; (3) sixth puff; and (4) seventh and eight puffs.
Data accessibility. This article does not contain any additional data. Authors' contributions. Y.Q. and X.S. conceived the study and designed this experiment. C.L. and E'.L. carried out the research. J.Z. and Y.T. carried out the data collection and analyses. C.Z. and Y.L. coordinated the study and helped analyse the experiment data. All authors gave final approval for publication.