B.A.T. (Investments) Limited, Regents Park Road, Millbrook, Southampton, SO15 8TL, UK.
Clairet Scientific Limited, 17/18 Scirocco Close, Moulton Park Industrial Estate, Northampton, NN3 6AP, UK.
Sci Rep. 2023 Aug 19;13(1):13501. doi: 10.1038/s41598-023-40422-z.
Use of e-cigarettes is increasing, alongside an expanding variety of devices and e-liquids. To match this growth and in line with the expanding legal and regulatory requirements applicable to manufacturers of e-cigarettes (e.g. disclosure of list of ingredients and quantities thereof in a product), rapid methods for determining levels of the main e-liquid constituents-namely, propylene glycol (PG), vegetable glycerol (VG), water and nicotine-are needed. We have assessed the ability of near infrared (NIR) spectroscopy, coupled with partial least squares (PLS) regression, to predict the levels of these constituents in e-liquid formulations. Using NIR spectral data from a large set of reference e-liquids incorporating working concentration ranges, flavourings, and other ingredients, linear calibration models were established for PG, VG, water and nicotine (predicted vs theoretical values, all R > 0.995). The performance of these models was then evaluated on commercial e-liquids using NIR and compared to results obtained by gas chromatography (GC). A strong correlation was observed between NIR-predicted values and measured values for PG, VG and nicotine (all R > 0.955). There was less consistency between predicted and GC measured values for water due to the relatively high limit of quantification (LOQ) of the GC method (2.6% w/w) versus the e-liquid content (0-18% w/w). The LOQ of the NIR method for water was 0.6% w/w, suggesting that NIR may be a more accurate method than GC to predict water concentration in e-liquids, especially at low levels (< 2.6% w/w). Collectively, although limitations of the technique have been identified, specifically for e-liquids containing compounds that might interfere with the set calibrations, our findings suggest that NIR combined with PLS regression is a suitable tool for rapid, simultaneous and high-throughput measurement of PG, VG, water and nicotine levels in most commercial e-liquids.
电子烟的使用正在增加，同时设备和电子烟液的种类也在不断扩大。为了适应这种增长，并符合适用于电子烟制造商的不断扩大的法律和监管要求（例如，披露产品中成分及其数量的清单），需要快速确定主要电子烟液成分的水平，即丙二醇（PG）、蔬菜甘油（VG）、水和尼古丁。我们评估了近红外（NIR）光谱结合偏最小二乘（PLS）回归来预测电子烟液配方中这些成分水平的能力。使用包含工作浓度范围、调味剂和其他成分的大量参考电子烟液的 NIR 光谱数据，为 PG、VG、水和尼古丁建立了线性校准模型（预测值与理论值，所有 R>0.995）。然后使用 NIR 和气相色谱（GC）评估这些模型在商业电子烟液上的性能，并将结果与 GC 结果进行比较。NIR 预测值与 PG、VG 和尼古丁的实测值之间存在很强的相关性（所有 R>0.955）。由于 GC 方法（2.6%w/w）与电子烟液含量（0-18%w/w）相比具有相对较高的定量限（LOQ），因此水的预测值与 GC 实测值之间的一致性较差。NIR 法测定水的 LOQ 为 0.6%w/w，这表明 NIR 可能是一种比 GC 更准确的方法，可以预测电子烟液中的水分浓度，特别是在低水平（<2.6%w/w）下。总的来说，尽管已经确定了该技术的局限性，特别是对于可能干扰设定校准的含有化合物的电子烟液，但我们的研究结果表明，NIR 结合 PLS 回归是一种快速、同时、高通量测量大多数商业电子烟液中 PG、VG、水和尼古丁水平的合适工具。
