Thompson Bruce T, Mizaikoff Boris
Brown & Williamson Tobacco Corp., P.O. Box 1056, Macon, Georgia 31202-1056, USA.
Appl Spectrosc. 2006 Mar;60(3):272-8. doi: 10.1366/000370206776342616.
Fourier transform infrared (FT-IR) spectroscopy was compared directly to independent standard analytical techniques for the routine measurement of carbon monoxide (CO) and nitric oxide (NO) yields from cigarette sidestream smoke. The FT-IR instrument was configured in-line with a nondispersive infrared (NDIR) analyzer for CO analysis and a chemiluminescence (CL) analyzer for NO analysis to monitor the sidestream smoke from a single port of a linear smoking machine. A cold trap was inserted prior to the FT-IR to minimize the levels of vapor phase interferents, such as water. Univariate and multivariate regression analysis were evaluated for the prediction of cigarette yield from time-resolved spectral data at 1, 2, 4 and 8 cm-1 spectral resolution. Regressions were developed using three different spectral ranges including unique rotation-vibration lines, the R-branch, and the entire absorption band. As per standard methods, yields were calculated from the concentration traces generated during the smoke runs for five different cigarettes spanning the expected range of mainstream total particulate matter deliveries. The FT-IR traces for the smoke runs revealed improved temporal resolution yielding analytical information from smoke generated in between puffs. The performance between the validation methods and the FT-IR calibrations was statistically compared. In general, for the determination of CO, the FT-IR calibrations underestimated the yield measured by NDIR by less than 10%. For the NO measurement, the univariate FT-IR calibrations overestimated the NO yield measured by the CL analyzer, whereas the partial least squares (PLS) calibrations showed good agreement. PLS calibrations were developed for both analytes providing no significant difference when compared to the respective standard analytical techniques. Results for sidestream CO and NO yields for Kentucky reference cigarette 1R4F utilizing 8 cm-1 calibrations compared favorably to values reported elsewhere in the literature. Hence, calibration of the FT-IR system at 8 cm-1 spectral resolution clearly revealed the potential of this method, providing enhanced temporal resolution, simultaneous determination of several smoke components, and reduced complexity of the experimental setup in contrast to the standard techniques.
将傅里叶变换红外（FT-IR）光谱法直接与独立的标准分析技术进行比较，用于常规测量卷烟侧流烟气中的一氧化碳（CO）和一氧化氮（NO）产量。FT-IR仪器与用于CO分析的非色散红外（NDIR）分析仪和用于NO分析的化学发光（CL）分析仪串联配置，以监测线性吸烟机单个端口的侧流烟气。在FT-IR之前插入一个冷阱，以尽量减少气相干扰物（如水）的水平。对单变量和多变量回归分析进行了评估，以根据1、2、4和8 cm-1光谱分辨率下的时间分辨光谱数据预测卷烟产量。使用三个不同的光谱范围进行回归分析，包括独特的转动-振动线、R分支和整个吸收带。按照标准方法，从五支不同卷烟的烟雾运行过程中产生的浓度曲线计算产量，这些卷烟涵盖了主流总颗粒物输送量的预期范围。烟雾运行的FT-IR曲线显示出时间分辨率的提高，能够从抽吸之间产生的烟雾中获取分析信息。对验证方法和FT-IR校准之间的性能进行了统计比较。一般来说，对于CO的测定，FT-IR校准低估了NDIR测量的产量，低估幅度小于10%。对于NO测量，单变量FT-IR校准高估了CL分析仪测量的NO产量，而偏最小二乘（PLS）校准显示出良好的一致性。为两种分析物开发了PLS校准，与各自的标准分析技术相比没有显著差异。使用8 cm-1校准的肯塔基参考卷烟1R4F的侧流CO和NO产量结果与文献中其他地方报道的值相比具有优势。因此，在8 cm-1光谱分辨率下对FT-IR系统进行校准清楚地揭示了该方法的潜力，与标准技术相比，它提供了更高的时间分辨率、同时测定多种烟雾成分以及降低了实验装置的复杂性。
