Table 1. Estimates of precision from the automated deter- Automated determination of mination of hydrogen cyanide, acrolein and total aldehydes in standard aqueous solution and in methanolic extracts of hydrogen cyanide acrolein the gas phase from a Canadian reference cigarette. and total aldehydes in the gas phase of tobacco smoke Concentration Precision* Cigarette Deliver_y ndminal recorder (relative (relative value response standard mean standard W.S. Rickert and P.B. Stockwell* Standard (ppm) (mm) deviation,%) (ug) deviation,%) Labstat Incorporated, 262 Manitou Drive, Kitchener, Acetaldehyde 60.0 25 6.8 On tario, Canada, N2C 1L3. 1,20.0 51 8.1 180.0 76 8.8 For some time there has been wordwide concern for the Hydrogen 5.0 36 3.4 health risks of cigarette smoking. In several countries, Cyanide 8.0 58 2.6 160_+14 including the United States, Canada and the United 10.0 72 3.3 surveys nicotine delivered by Kingdom, of the tar and Acrolein 5.0 37 6.1 58_+4.6 commercial brands of cigarettes are carried out and published. 8.0 58 5.0 Recently carbon monoxide levels have also been measured. 10.0 73 2.9 The combustion of tobacco products such as cigarettes also produces three major ciliatoxic components, namely hydrogen cyanide (HCN), formaldehyde and acrolein ]. As * 120 observations in the case of nicotine and carbon the + 120 observations on 'Monitor D' over a period of 3 months. tar, monoxide, A detection of statistically significant differences requires the Confidence interval for 10 observations (P> .95) results of from 10 to 30 analyses per brand per chemical parameter at several points in time; as a consequence, the use Table 2. The effect of trapping solution on the recovery of automated chemical analyses is advantageous. Individual of selected cigarette gas phase constituents. automated colorimetric methods have been published for HCN, acrolein and total aldehydes using AutoAnalyser II techniques; however, the simultaneous determination of all Constituent (/ag/cigarette)* three from a single sample has not been attempted due, in A part, to the fact that different gas scrubbing agents and Solvent HCN ALDEHYDES ACROLEIN diluents must be used for each analyte. Selection of a single efficient scrubbing agent and the introduction of on-line Water 81 +_5.5 718 _+ 25 18_+ 1.5 dilution facilitate the simultaneous determination of the amount of HCN, acrolein and total aldehydes (mainly Methanol 99_+ 7.1 830 _+ 20 57 _+4.6 acetaldehyde) delivered by various brands of cigarettes. This improves the overall laboratory efficiency and minimises the Ethanol 95 +_ 7.2 850 +_ 39 52 -+3.8 manual effort in the analytical methodology. Equipment * 10 observations on a Canadian reference cigarette were obtained. In the present procedure the cigarette smoke is automatic- ally generated using a Phipps and Bird 20-port smoking AThe choice of solvents was based on the work of others who have machine. After conditioning for 48 hours at 60% relative used water [7] methanol [4] and ethanol [5] for extracting aldehydes, and acrolein in similar applications. humidity and 22 each cigarette is lit and a puff of 35 ml HCN, respectively, C Temperature was maintained at 2 +_ 2 C using a circulating bath and is taken once per minute during 2 second intervals until a 50% ethylene glycol. butt length of 30mm is reached. Five cigarettes of a particular brand are smoked for each analysis.The particulate matter is trapped on a filter pad while the gas phase is passed enylhydrazine to form substituted hydrazones which in turn through a sintered glass frit fitted into a trap behind the produce coloured complexes with sodium hydroxide; cigarette holder. The traps contain 35 ml of a suitable solvent acrolein is determined using the method developed by Cohen (water, methanol or ethanol) thermostatted to 2 C using and Altshuller [5] whereby, in the presence of mercuric a circulating bath containing 50% ethylene glycol and 50% chloride and trichloracetic acid, acrolein condenses with water Immediately upon completion of the smoking 'run' 4-hexylresorcinol to give a blue complex. All three reactions of five cigarettes per sample, a 4 ml sample from each of the are monitored colorimetrically. twenty traps is transferred into a sample cup which is sealed with a stretched piece of plastic film ensuring that volatile Procedure aldehydes are not lost whilst the samples are on the turn- Figure is a representation of the manifold used for simult- table awaiting analysis.The AutoAnalyserequipment used for aneously measuring the three analytes. The twin sampling the analyses consists of a dual probe sampler, a proportion- probe incorporated in the sampler is used to deliver the ing peristaltic pump (26 channel), three colorimeters, a three- sample stream to the manifold. One probe delivers the pen recorder and a data processing system. The reactions sample for the acrolein analysis whilst the other supplies a employed are well established [2, 3, 4]: the determination two-way stream splitter from which a sample is drawn for of HCN is based on the Konnig reaction whereby cyanogen determination of HCN and total aldehydes. On-line mixing chloride (obtained from the reaction of chloramine T with and sub-sampling must be carried out effectively and repro- HCN) reacts with pyridine to produce glutaconic aldehyde ducibly. Simple use of coloured solutions enables the which in turn forms a coloured complex with pyrazolone; efficiency of the mixing assembly, shown in Figure 2, to be aldehydes are determined by reaction with 2, 4-dinitroph- checked visually. Since there is a tendency for air introduced on sample switching to become trapped in the sub-sampling * Present address: Laboratory of the Government Chemist, Cornwall well, bubbles must be removed prior to this point as illus- House, Stamford Street, London, SE1 9NQ, UK. trated in figure 2.. The balance of flow rates used for the 152 The Journal of Automatic Chemistry Rickert & Stoekwell Automated determination of hydrogen cyanide etc. FLOW RATE (ML/MIN) A. HYDROGEN CYANIDE WAS TE WASTE .60 DEBUBBLE 0.80 WASTE BUFFER I. 20 AIR 0.80 lOT CHLORAMINE T (S) 0.:30 PYRAZOLONE REAGENT (A) 2.90 540 mm SAMPLE (S) 0.42 WASH (WATER) 2.50 ) WASH DILUENT (0.1N N(:IOH (,S) 2.00 SUBSAMPLE (S) 0.42 S TE 6WA-0202-05K 116-0200-04 B. TOTAL ALDEHYDES lOT WAS SAMPLE (S) 0.42 28T 20T METHANOL NOH (S) 0.88 lOT AIR 0.80 480mm 2-4 DN P (A) 0.42 2N N(OH (A) 1.20 'VASTE WASTE (S) I. 40 DILUENT ,50 % MeOH) (S) 1.20 SUBSAMPLE (S) O. 8 DEBUBBLE 0.80 WASTE C. ACROLEIN --- WASTE) SAM PLE (S) 1.60 0 lOT lOT lOT TRICHLORACETEC ACID (A) 2.34 0 r. AIR 20 MERCURIC CHLORIDE (S) 0.25 "'-.- HEXYLRESORCINOL (S) 0.25 ,-," 40T SAMPLER WASH 1.60 0 =-- WASH WASTE (A) .zo ( - WASTE (S) SOLVAFLEX (A) ACIDFLEX Figure 1. Flow diagram of the analytical system used to determine the hydrogen cyanide acrolein and total aldehyde contents of a methanolic tobacco smoke extract. sample and dilution stream and the solubility and density of cigarette. Although the individual methods are simple adapt- the stream are important parameters which must be matched ations of the published procedures the integrated system as for the reactions involved and the inclusion of a mixing illustrated is unique in utilising a single sample with on-line c0il just after diluent addition helps in this regard. For HCN dilution and sub-sampling. In this way, maximum responses the linear absorbance range of the colorimeter is 0 to 0.45 are obtained and the anticipated levels of HCN and volatile 0D which corresponds to a level of 0 to 13 mg/kg. A similar aldehydes are reduced to within the range required for range is also obtained for acrolein whilst the level for total linear colorimetric response. aldehydes is 0 to 0.35 OD corresponding to 0 to 230 mg/kg. A suitable standard solution for each of these analytes is used to adjust the data processing equipment which provides Discussion direct printout in concentration units. Once calibrated, As already stated, the ability of the system to produce the samples generated from the smoking machine can be reliable determinations depends on efficient mixing and analysed simultaneously for HCN, acrolein and total stable reaction chemistries. Table is a summary of the perf- aldehydes with the results printed in the units /2g "per ormance of the system under realistic conditions over several Volume No. 3 April 1979 153 Rickert & Stockwell Automated determination of hydrogen cyanide etc. TO WASTE (DEBUBBLER) SLEEVING 116-0202- 05K SAMPLE 116-0200-04 DI LUENT STAINLESS STEEL TUBING ----- TRANSMISSION ING (0.03" I.D.) MIXING COIL- lOT, ALDEHYDES WASTE SUBSAM PLE ST, HCN ----- TO MANIFOLD Figure 2. The on-line dilution system. months. With respect to the standard solutions, at first glance REFERENCES the precision appears to be somewhat greater than the level [1 Battista, S.P. in 'Proceedings 3rd World Conference Smoking and of 2 4% reported by others for analyses of this type [6]. Health' Ed. Wynder, E.L., Hoffman, D., and Gori, G.B., 1975, However, it must be remembered that estimates of variance US Department of Health Education and Welfare Publication include day-to-day variations in colorimeter response as well No. (N1H) 76-1221p 517. Collins, P.F., Sarji, N.M., and Williams, J.F., Be#rage Zur Tabak- as [2 week-to-week variations due to standard and reagent forschung, 197 3, 7, 73. preparation. The variability in the yields from the 'monitor' [3] Collins, P.F., Sarji, N.M. and Williams, J.F., Tobacco Science, cigarette are still larger due to the additional component of 1970, 14,182.