Tobacco products were purchased from online retail outlets in the U.S. or from commercial establishments in the greater Atlanta, GA area between 2014 and 2016. Samples were assigned unique identifiers and were logged in to our LIMS data base when received. Samples were stored in their original packaging until analyzed. Only authorized lab personnel had access to the samples.

Tobacco samples from commercial cigarettes, Quality Control moist snuff tobacco samples 1S3 and CRP2 (North Carolina State University, Raleigh, NC, USA) little cigar filler tobacco, and most smokeless tobacco products were dried for a minimum of 1 hour at 90°C. Syrupy moist smokeless tobacco products were dried for up to 16 hours to render moisture contribution to sample mass negligible. Dried tobacco was rendered more homogeneous by grinding for 20 seconds with a Smart Grind coffee grinder (Black and Decker, Middleton, WI, USA).

Dried tobacco samples (0.100 to 0.125 g) were digested in TFM vessels with 9 mL environmental grade nitric acid (GFS, Powell, OH, USA) that was further purified by distillation in a perfluoroalkoxy (PFA) sub-boiling still (CEM, Matthews NC, USA), 2 mL 30–35% double distilled hydrochloric acid, and 0.5 mL 46–51% double distilled hydrofluoric acid (GFS, Powell, OH, USA) to assure complete dissolution of silica and aluminum silicates. Procedural blanks were prepared by adding the same digestion reagents to a TFM digestion vessel. The procedural blanks were then processed through the microwave digestion and dilution procedures as if they were samples. Sample digestion was accomplished by a 10 minute ramp from ambient temperature to 190°C, and maintaining 190°C for 15 minutes with a Milestone Ethos microwave system (Shelton, CT, USA). Digested samples were rinsed from digestion vessels into acid-cleaned 50 mL polymethylpentene (PMP) class A volumetric flasks and diluted to volume with ultrapure water (>18 MΩ·cm). 5.00 mL aliquots of the diluted digestates were further diluted with 1% v/v nitric acid and 1% v/v hydrochloric acid to 100 mL in PMP volumetric flasks.

Analyses of digested tobacco were performed using an Element XR Sector Field Inductively Coupled Plasma-Mass Spectrometer (ThermoFisher, Bremen, Germany). Aliquots (10 mL) of calibration blanks, standards, procedural blanks, QCs and diluted samples were transferred to acid-cleaned 15 mL “Metal Free” polypropylene sample tubes (VWR, Atlanta, GA, USA) and placed in a FAST autosampler (ESI, Omaha, NE, USA) for analysis. The sample introduction system included a PFA-ST nebulizer, Peltier-cooled PC3 PFA spray chamber, o-ring free sapphire injector and high performance torch (ESI). Samples were teed with internal standard 10 μg/L gallium in 1% v/v nitric acid and 1% v/v hydrochloric acid ahead of the nebulizer and pumped simultaneously at 17 rpm through 0.76 mm i.d. peristaltic pump tubing to the nebulizer for approximately 600 µL/minute liquid flow rate. Plasma gas flow was 16 L/min with 1210 Watts forward power. Sample gas and auxiliary gas were optimized for highest possible stable aluminum signal while maintaining oxide formation at 1% or less, generally near 1.10 L/minute with auxiliary gas near 0.80 L/minute. Other parameters were optimized for optimum signal. Platinum-tipped nickel sampler and H skimmer cones were obtained from Spectron (Ventura, CA, USA). Data was acquired in medium resolution (r ≥ 4,000, 10% valley definition) with 3 runs and 10 passes, 0.010 second sample time, 20 samples per peak in Mass Accuracy mode using auto-lock mass with mass offsets determined for ²⁷Al and ⁶⁹Ga (internal standard) after mass calibration. Mass windows were 100%, with 50% search windows and 60% integration windows.

Instrument calibration was established with High Purity Standards aluminum standard 10001–1 (Charleston, SC, USA) diluted into 1% v/v nitric acid and 1% v/v hydrochloric acid to prepare a 5 calibration standard range from 12.5 to 300 µg/L. Assuming a minimum of 100 mg dried tobacco per digestion, the calibrated reportable range is from 0.125 µg/mg tobacco to 3.00 µg/mg tobacco. The calibration blank consisted of the acid solution in which calibration standards had been prepared. Gallium internal standard (10 µg/L, NIST 3119a, Gaithersburg, MD, USA,) was prepared in the same acid solution. The calibration blank and standards were analyzed at the beginning of each analytical run. Calibration was considered acceptable if the correlation coefficients were ≥ 0.999. The internal standard corrected calibration blank background was subtracted from calibration standards. The procedural digestion blank was subtracted from quality controls (QCs) and samples.

Results obtained from twenty duplicate analyses of Smokeless Tobacco Reference Products 1S3 and CRP2 used as tobacco quality control samples and quintuplicate run results from Standard Reference Materials BCR-482 (lichen), NIST 1570a spinach Leaves, and NIST 1573a tomato leaves) were used to calculate a plant matrix specific method Limit of Detection (LOD) using a Taylor plot of standard deviations versus mean results to determine S₀ (23). Three times the standard deviation obtained by extrapolation of the regression line to 0 µg/mg concentration was used to calculate the method LOD. The results of the analyses of the same reference products were also used to determine precision and accuracy of the method.

Quality control was maintained by analysis of Smokeless Reference Tobacco Product (STRP) 1S3 and CRP2 before and after each group of samples. The analytical QC samples were evaluated using a modified Westgard evaluation approach (24). When a QC was determined to be out of control according to the modified Westgard criteria, results in the respective batch were not used and analyses were repeated.

The calculated LOD for tobacco aluminum was 0.012 µg/mg dried tobacco. Only concentrations greater than the lowest calibration standard concentration (12.5 µg/L) divided by 100 mg, the lower sample mass limit for sample preparation, were reported as greater than the Lowest Reportable Level (LRL), providing a more conservative lower limit for reportable concentrations. The LRL based on the lowest calibration standard was 0.125 µg/mg tobacco, ten times higher than the calculated LOD. Similar calculation for the highest calibration standard resulted in a 3.00 µg/mg tobacco Highest Reportable Level (HRL). All sample concentrations were greater than the LRL and lower than the HRL.

Precision based on five analyses of QCs, CRMs, STRPs, and SRMs was ± 11.7% at 0.568 µg Al / mg leaf material for NIST SRM 1573a and ± 3.2% at 0.819 µg/mg for STRP 1S3. Acceptable accuracy was validated based on comparison of analytical results at two concentration levels with certified aluminum concentrations in SRMs (Table 1).

The results from quintuplicate analyses of smokeless tobacco, little cigar, roll your own/pipe tobacco, and cigarette tobacco for aluminum concentrations are shown in Table 2. Based on the inclusion of cigarette, little cigar, pipe, roll-your-own tobacco (which constitute the majority of “any combustible tobacco product” category), as well as smokeless tobacco products in this study, these samples are representative of approximately 98% of tobacco products used by over 60,000 respondents in the United States National Adult Tobacco Survey, 2013–2014 (25). All sample concentration results were within the reportable range. The tobacco aluminum concentrations were similar to those reported for a more limited range of products (22). It should be noted that the concentrations are “total aluminum,” not speciated as soluble or insoluble forms.

Though the sample number is small, it is apparent that as a product group, commercial smokeless tobacco samples had the lowest aluminum concentrations. Though there was some overlap between the other product groups, the mean aluminum concentrations and concentration ranges in little cigar and roll your own/pipe tobacco group were higher than in cigarette tobacco.

In addition to possible differences in soil preparation for crops that were used for the different products, another possible reason for lower aluminum concentrations in the smokeless tobacco group could be product preparation. Many smokeless, cigarette, roll your own/pipe tobacco, and little cigar tobacco products are cut into strips or finer pieces as evident from examination of the products. The fines resulting from cutting are recovered as reconstituted tobacco sheet that is mixed with filler for many cigarettes, and used to manufacture wrapping paper for many little cigars. Reconstituted tobacco is not typically a component of traditional smokeless tobacco products (26). We have previously shown that superficial aluminum silicates from soil are visibly higher in reconstituted tobacco than on smokeless or cigarette filler tobacco surfaces (18). It is possible that many superficial aluminum silicate particles are dislodged as fines during cutting and recovered along with tobacco leaf fines in reconstituted sheet. This is one possible reason that the small number of loose leaf (Southern Pride, Taylor’s Pride) and moist snuff (Copenhagen, Red Seal) smokeless tobacco products tested in this study collectively had lower total aluminum concentrations than the other types of tobacco products.

We have previously shown using scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS), that aluminum silicate particles are present in mainstream smoke (17). Insoluble aluminum silicate particles have been described as “smokers’ inclusions,” and as the principal particles visible in pulmonary macrophages of smokers (14, 15, 16). While insoluble particles such as kaolin and halloysite aluminum silicates can be imaged using SEM-EDS, soluble forms of aluminum dispersed in the leaf tissue would not be visible. Therefore, it is likely that a substantial portion of the total aluminum transferred from tobacco into mainstream smoke that is inhaled by smokers is in insoluble form, although a small amount may be transported in soluble form.

Whether aluminum in mainstream tobacco smoke is in soluble or insoluble form determines the mechanism of toxicity. In soluble form, aluminum could be taken up by pinocytosis at the pulmonary epithelium, or by other mechanisms. Ultimately, aluminum in soluble form could enter circulation, accumulate in tissues, and exert toxic effects including neurotoxicity effects (2). Insoluble forms greater than 100 nm diameter would generally be ingested by macrophages and induce inflammatory response, whereas insoluble nanoparticles smaller than 100 nm could be taken up by bronchoalveolar macrophages, or enter interstitial macrophages and other cells and cause inflammation (20, 21, 27, 28). Therefore, further work is necessary to characterize the proportions of aluminum in tobacco and mainstream smoke that might be in soluble form, and the predominant size ranges of the insoluble forms.