Comprehensive characterization of mainstream marijuana and tobacco smoke

Recent increases in marijuana use and legalization without adequate knowledge of the risks necessitate the characterization of the billions of nanoparticles contained in each puff of smoke. Tobacco smoke offers a benchmark given that it has been extensively studied. Tobacco and marijuana smoke particles are quantitatively similar in volatility, shape, density and number concentration, albeit with differences in size, total mass and chemical composition. Particles from marijuana smoke are on average 29% larger in mobility diameter than particles from tobacco smoke and contain 3.4× more total mass. New measurements of semi-volatile fractions determine over 97% of the mass and volume of the particles from either smoke source are comprised of semi-volatile compounds. For tobacco and marijuana smoke, respectively, 4350 and 2575 different compounds are detected, of which, 670 and 536 (231 in common) are tentatively identified, and of these, 173 and 110 different compounds (69 in common) are known to cause negative health effects through carcinogenic, mutagenic, teratogenic, or other toxic mechanisms. This study demonstrates striking similarities between marijuana and tobacco smoke in terms of their physical and chemical properties.

Health Canada Intense (HCI) puffing parameters (55 mL puff of 2 s duration, every 30 s, for 8 puffs) 16 . HCI was selected as it is a compromise between the ISO standard 17 puff routine (35 mL puff of 2 s duration, every 60 s, for 8 puffs) commonly used for tobacco cigarette smoking and the more aggressive routine (70 mL puff of 2 s duration, every 30 s, for 8 puffs) developed by Moir et al. 14 to represent the longer/deeper inhalation techniques used during marijuana joint smoking 9-12, 14, 15 .
For aerosol characterization, mainstream smoke was sampled from the cigarette or joint through the heated puff head into a cylindrical lung bag (6.4 cm diameter and 35 cm length). Immediately prior to starting the puff routine, the lung bag was pre-filled using the heated breath head with 300 mL of HEPA-filtered air to act as a buffer volume during smoke generation. To limit smoke particle coagulation during the puff routine, 297 mL of HEPA-filtered dilution air was also added to the lung bag immediately prior to each 55 mL puff. The 355 mL volume (i.e. combined smoke and dilution air volume added) present in the lung bag after each puff was transferred through the heated breath head into a cylindrical sample bag (29.1 cm diameter and 100 cm length) between each puff. At the end of the puff routine, the remaining volume in the lung bag (i.e. last puff with dilution air and buffer air) was transferred to the sample bag. Prior to connecting to the SCS, the sample bag was manually pre-filled with 30 L of HEPA-filtered dilution air to provide sufficient volume for sampling with the particle characterization equipment. The lung and sample bag were both replaced between each cigarette or joint smoked, and were made of medium-duty (62.5 micron thick), polyethylene tubing (Kite Packing, Coventry, UK) with one end heat sealed. To avoid cross-contamination, the smoke generation apparatus was cleaned at the start of each day or when switching between smoking tobacco or marijuana samples.
For chemical analyses and total particulate matter (TPM) measurements, the non-vapor component of the smoke produced was sampled using quartz filters which were mounted in a stainless steel filter holder attached to the heated SCS puff head. This setup captured mainstream smoke directly from the mouth end of the cigarette or joint, therefore limiting sample losses and phase re-partitioning. During this filter loading, the stainless filter holder was cleaned between each cigarette or joint smoked and no lung and sample bag were required.
The filters (for chemical analyses and TPM measurements), cigarettes and joints were conditioned prior to sampling at a temperature of 22±1 • C and relative humidity of 60±3% for a minimum of 48 hours, in line with the ISO standard for tobacco conditioning and testing 18 . Flameless ignition was achieved using a coiled electric heater with a flat surface.

S3 Aerosol analyses
To assess some of the effects of the steady-state sampling on the smoke characteristics, the diffusion and settling losses of the particles during the aerosol analyses were estimated. The particle losses in each aerosol instrument was considered within its standard inversion procedure (i.e. SMPS correction following TSI 19 or AAC correction following Johnson et al. 20 ) or accounted for based on a previous study (i.e. catalytic stripper correction following Dickau et al. 21 ). The other aerosol measurements (i.e. particle mass, effective density and semi-volatile fraction) did not require corrections for particle losses in the DMA, CPMA or catalytic stripper as the results only depended on the location of the measurement peak in the mobility or mass domain, not its amplitude.
The particle losses in the sample bag were estimated following a similar approach as Johnson et al. 22 , which is based on the aerosol theory summarized by Hinds 23 . The diffusion and losses in the sample bag were found to be insignificant to the lognormal parameters (< 5.5% change) of the smoke size distributions. As expected, these corrections slightly increased the amplitudes of both the lower (i.e. diffusion losses) and upper (i.e. settling losses) extremities of the size distributions. As a result, the GSD and N for both the mobility and aerodynamic size distributions increased slightly, between 0.3-1.0% and 2.2-3.7%, respectively, for all but one smoke sample. The nonstripped marijuana smoke, with its larger particles, had slightly higher settling losses in the sample bag with the GSD and N of both the mobility and aerodynamic size distributions increasing Both CPCs shown in Fig. S2 and used in the other experimental setups for the aerosol portion of this study were the same model (TSI Inc., Shoreview, MN, USA: Model 3776) and detect particles with diameters from 2.5 nm to 3 µm 28 . Since the particle number concentrations in the sample bag decreased over time and the AAC scans were approximately 15 mins long, the particle number concentrations upstream of the AAC (measured by CPC 1) were used to normalize those downstream of the AAC (measured by CPC 2). This correction method was also utilized by Johnson et al. 22 for effective density measurements of aged tobacco smoke particles.
To measure the mobility size distributions of the aerosol samples, a Scanning Mobility Particle Sizer was used (SMPS; TSI Inc.: 3077 Bipolar charge conditioner, 3080 Electrostatic Classifier, 3081 Differential Mobility Analyzer [DMA], 3776 Condensation Particle Counter [CPC]). A bipolar charge conditioner was placed upstream of the DMA to charge the aerosol to a known equilibrium distribution of electrical charges. The DMA uses an electrostatic force to move the charged particles that are in-turn resisted by their drag force in the carrier gas. Only particles with a narrow range of mobility (-equivalent) diameters follow the correct trajectory and pass through the DMA 29 . This mobility diameter setpoint § can be set anywhere between 10 nm and 1 µm by varying the classifier voltage and gas flows of the DMA 19 . At the flow rates used in this study, 0.3 and 3 L min −1 for the sample and sheath flow ‡ , respectively, the classification range of the DMA operating within an SMPS over a 120 second scan was approximately 15 to 685 nm. By measuring the classified particle concentrations using a CPC (labelled CPC 2 in Fig. S2) as the DMA setpoint is scanned, the mobility size distribution of an aerosol can be quantified 30 . This measured distribution was corrected for multiply-charged particles following He & Dhaniyala 31 . The five consecutive mobility size distribution measurements (i.e. consecutive three minute scans started after the puff routine finished) were repeated three times, each from a new sample bag, and averaged to generate the size distribution parameters reported in Fig. 1b and in-text. The mobility scan shown in Fig. 1b is an average of all fifteen SMPS scans, accounting for the change in the smoke particle concentration over time by scaling the entire distribution by its total particle concentration relative to that from the first scan.
Both the aerodynamic and mobility particle size distributions were characterized by a log-normal distribution using a least-squares regression to determine the fitted count median diameter (CMD), geometric standard deviation (GSD), and total number concentration. These measurements focused on the measured CMD and GSD of the smoke particles, as the total particle number concentrations quantified by these methods were limited due to the transient nature of the smoke particles as discussed in the next section.

S3.2 Measurement of total number concentration
The particle number concentration of each aerosol sample was directly measured as a function of time using CPC 1 as labelled in Fig. S2. The particle number concentration was also indirectly measured by integrating the area under the aerodynamic or mobility size distributions.
The measured particle number concentration decreased over time as shown in Fig. S3 and Fig. S4. These concentrations are corrected for the sample dilution factor of 75 which accounts for the 30 L of dilution air in the sample bag, and 297 mL of dilution air per puff and 300 mL of buffer dilution air added into the lung bag. Fig. S4 shows the mobility size distributions of the nonstripped tobacco smoke particles measured by the SMPS as a function of time. This measurement demonstrates that the mobility CMD remained relatively constant (within 1.4% of the average CMD of nonstripped tobacco smoke) between the 5 consecutive 3 minute scans (i.e. 15 minute total measurement time). The change in the mobility CMD over the 5 consecutive mobility scans from either smoke source and condition (nonstripped vs stripped) varied by less than 5.4% over a similar period.
All three measurement methods of particle concentration varied due to the transient nature and high concentration of the smoke particles. The reported CPC 1 measurements include a coincidence correction as the measured particle concentrations were greater than 3×10 5 particles cm −3 , a threshold above which the CPC has reduced measurement accuracy 28 . Furthermore, CPC 1 was measuring at the maximum of its range (1×10 6 particles cm −3 ) for the first 0 to 15 seconds of sampling depending on the smoke source and its conditioning (nonstripped or stripped). It should be noted that these conditions did not occur in CPC 2 as the particle concentrations were below the CPC coincidence correction threshold after classification by the AAC, DMA, or CPMA. The number concentrations determined from the mobility measurements are sensitive to the SMPS multiple-charge correction 20,32 . Furthermore, the common practice for characterizing steady-state aerosols (as alluded to in Annex G of BS ISO 15900 33 ) is to use a SMPS to measure the mobility size distribution and in parallel use a CPC to measure the total particle number concentration since charge corrections are not required for a CPC. While the aerodynamic measurements do not require a multiple charge correction 20 , its long scan time (≈15 minutes) of a transient aerosol likely had the greatest effect on the area under the measured aerodynamic size distribution. This inference is supported by the stability of the mobility CMD between consecutive SMPS measurements. However, despite these additional sources of concentration measurement errors, the particle § Assuming the particles are singly charged.  concentration ratio between marijuana and tobacco smoke from all three measurement methods agreed within 20% for both nonstripped and stripped aged smoke samples as shown in Fig. 1c and in-text.

S3.3 Measurement of effective density
Effective density of nonstripped smoke particles as a function of particle mobility diameter was measured using the setup shown in Fig. S5a. Similar to the experimental setups for aerosol size distribution measurements, a sample flow rate of 0.6 L min −1 was used. However, in this case, the smoke particles were first classified by mobility using DMA 1 at a constant diameter setpoint prior to a second aerosol classifier. Similar to the previous setups, the sample flow was split, with 0.3 L min −1 drawn by CPC 1 to measure the total number concentration of the classified particles (i.e. downstream of DMA 1), which were monodispersed in electrical mobility. The remaining 0.3 L min −1 classified sample flow was passed through a centrifugal particle mass analyzer (CPMA) to classify the particles by their mass-to-charge ratio using opposing centrifugal and electrostatic forces generating by charged, rotating concentric cylinders 34 . This mass-to-charge setpoint § can be set anywhere between 0.2 ag and 1050 fg by varying the classifier speed and voltage of the CPMA 35 . At the sample flow of 0.3 L min −1 and the classification resolution of 20 used in this study, the classification range of the CPMA was approximately 3.8 ag to 296 fg. The CPMA was operated at twice the classification resolution of the AAC or DMA (i.e. 20 vs 10) to further distinguish the charge states of the particles classified by the upstream DMA. By measuring the classified particle concentrations using a CPC (labelled CPC 2 in Fig. S5a) as the CPMA setpoint is stepped, the mass-to-charge distribution of an aerosol can be quantified 36,37 . Since the charge states of the particles are known from the upstream neutralizer and DMA, the particle mass classified by the CPMA at the given DMA setpoint can be determined. Using the measured particle mass and mobility diameter, the effective density of the particles can be calculated following Eqn. 1 of main-text.
The effective densities of stripped particles were measured using the setup shown in Fig. S5b. The 0.6 L min −1 sample classified by DMA 1, which was monodispersed in particle mobility, was passed through a catalytic stripper at 350 • C. The 6/95 effective density of the particles was measured by using the two downstream classifiers, DMA 2 and the CPMA, in parallel to measure the mobility diameter and mass of the stripped particles, respectively. Each downstream classifier was operated in tandem with a CPC operating at at 0.3 L min −1 . Due to this tandem classifier arrangement, DMA 2 was not operated in scanning mode like an SMPS to avoid introducing additional errors into its mobility measurements. These errors are caused by the upstream DMA (i.e. DMA 1) producing a narrow mobility distribution, which invalidates the SMPS inversion assumption that the aerosol distribution is constant over the width of the downstream DMA (i.e. DMA 2) transfer function 38 . To avoid this issue, the downstream DMA was stepped, rather than scanned, and the tandem DMA inversion developed by Stolzenburg & McMurry 39 was applied. This tandem DMA inversion was also previously used to study the hygroscopic properties of tobacco smoke particles 40 .
For both the nonstripped and stripped smoke samples, this process was repeated at various particle sizes (i.e. different DMA setpoint) to identify the relationship between particle mobility diameter and effective density as shown in Fig. 1d and in-text.

S3.4 Measurement of total particulate matter
The total particulate matter (TPM) produced by smoking a tobacco cigarette or marijuana joint was measured by collecting fresh smoke (i.e. without dilution or aging) on a quartz filter directly downstream of the cigarette or joint. The quartz filters (47 mm Type A/E Glass Fiber Filter from Pall Life Sciences [P/N 61631]) were weighed before and after loading using an ultra-microbalance (UMX2; Mettler Toledo, Columbus, Ohio, USA). To avoid the TPM filters becoming overloaded and causing the puff routine to deviate significantly, only six of the eight puffs of the HCI smoking routine were used during the TPM filter loading for both the tobacco and marijuana samples. A TPM sample was deemed acceptable if the total puff volume was within 10% of the 330 cm 3 target. The total puff volumes deviated by -0.6% to -3.2% for the TPM filters loaded with tobacco smoke, and by 0.2% to -8.7% for those loaded with marijuana smoke. These deviations are on the same order of magnitude as the uncertainty of other studies, such as the 5% to 7% confidence interval calculated from the data of Moir et al. 14 , which had a larger TPM sample size of 30 repeats.
Immediately after loading the filters were placed on the ultra-microbalance, however each filter's mass would continue to decrease for tens of minutes. This decrease was small relative to the total TPM (less than 3.6%) and is likely due to the volatility of the TPM. This observation at room temperature supports the conclusion that the particles from either smoke source have a highly volatile component. The TPM filters were also conditioned before and after loading, however the change in total mass due to conditioning after loading was insignificant (less than 0.6%). Over the 6 HCI puffs, the marijuana joint produced 27 (±4) mg of TPM, while the smoking the tobacco cigarette produced 7.8 (±0.9) mg of TPM. The reported TPM measurement limits represent the 95% confidence of the measurement repeatability assuming a t-distribution. These results agree with the results of Ingebrethsen et al. 41 , which found smoking the same 3R4F tobacco cigarette following the same HCI puff routine produced 1.0±0.3 mg and 1.9±0.3 mg of TPM for puff 2 and puff 5, respectively, thus agreeing with the 1.3±0.3 mg average TPM per tobacco cigarette puff currently measured. This per puff average is only a high-level comparison as it neglects that the TPM produced changes as a function of rod length or with each puff over a smoking routine.
Based on the TPM measurements and associated confidence intervals determined in the current study, smoking a marijuana joint produces roughly 3.4 (±0.6) times more TPM than a tobacco cigarette following the same six puff routine. This result qualitatively agrees with our observation that more than six HCI puffs from the marijuana joint would overload the TPM filter causing significant puff deviations, while no such overloading and puff deviations were observed for TPM samples loaded with up to eight HCI puffs from the tobacco cigarette. However, this 3.4 ratio is contrary to the results of Moir et al. 14 , which found tobacco cigarettes and marijuana joints produce approximately the same amount of TPM per puff following either the ISO (35 ml puff over 2 seconds every 60 seconds) or extreme (70 ml puff over 2 seconds every 30 seconds) smoking routines. This difference between the studies could be due the different puffs routines used, and that the tobacco cigarettes studied by Moir et al. 14 were nonfiltered as filtered cigarettes are known to reduce the TPM produced from smoking tobacco 5 . The inference is supported by another study 42 which found on average that mainstream smoke from a marijuana joint yields on average 2.2 times greater TPM than mainstream smoke from a filter-tipped tobacco cigarette.

S3.5 Measurement of semi-volatile mass and volume fractions
The semi-volatile mass and volume fractions at a given aerosol particle size were measured using the setup shown in Fig. S5a and S5b. The semi-volatile mass fraction at a given particle size was measured by comparing the masses of nonstripped and stripped particles measured by the CPMA at a given DMA setpoint, in this case 470 nm mobility diameter. Specifically, the count median mass (CMM) measured by the CPMA in Fig. S5a is compared with the CMM measured by the CPMA in Fig. S5b. Each CMM was determined by fitting a log-normal distribution to each charge peak of the CPMA scan using a least-squares regression.
The semi-volatile volume fraction at a given particle size was measured by comparing the mobility-equivalent volume of nonstripped and stripped particles measured by DMA 2 at a given DMA 1 setpoint as shown in Fig. S5b. Note that DMA 2 was operated in stepping mode for the reasons described in Section S3.3. Specifically, the TDMA inversion median diameter of the 7/95 stripped particles measured by DMA 2 is compared with the nonstripped particle size selected by DMA 1 (i.e. 470 nm mobility diameter setpoint).
The semi-volatile fractions for the polydispersed size distributions were calculated using the mass and volume concentrations of the aerosols, rather than the individual particle mass and volume. The volume and mass concentrations were estimated using the Hatch-Choate equations 23 and the lognormal distributions fitted to the measured mobility size distributions (as described in Sections S3.1). The fitted lognormal distribution was used to account for particles smaller or larger than the SMPS measurement range (14.6 to 685.4 nm). It is clearly shown in the mobility size distributions of Fig. 1b that particles larger than the SMPS measurement range are present. The mass concentration estimate also utilized the effective density of the particles as described in Section S3.3.

S4 Chemical analyses
Volatile compounds from particulates captured in pre-fired quartz filters were sampled by headspace solid-phase microextraction (HS-SPME) followed by comprehensive two-dimensional gas chromatography with a time-of-flight mass spectrometric detector (GC×GC-TOFMS). The pre-fired quartz filters (47 mm PallFlex Tissuquartz TM 2500 QAT-UP filters from Pall Life Sciences [P/N 7202]) utilized for the chemical samples were different than the TPM filters and appeared to generate a lower pressure drop during loading. As a result, the chemical sample filters did not cause the puffs to deviate late in the smoking routine, and thus could be consistently loaded using all eight puffs for both the tobacco and marijuana samples.

S4.1 HS-SPME-GCxGC-TOFMS
The GC×GC-TOFMS system consisted of an Agilent 7890 (Agilent Technologies, Palo Alto, CA, USA) gas chromatograph and a Pegasus 4D TOFMS (LECO, St. Joseph, MI, USA) with quad jet liquid nitrogen-cooled thermal modulator. The first dimension ( 1 D) column was a low-polarity 5% phenyl / 95% polydimehthylsiloxane-type phase (Rtx-5MS; 60 m × 0.25 mm i.d.; 0.25 µm film thickness) connected by means of a SilTite µ-Union (Trajan Scientific and Medical, Victoria, Australia) to a second dimension ( 2 D) mid-polarity trifluoropropylmethyl polysiloxane-type phase (Rxi-200; 1.6 m × 0.25 mm i.d.; 0.25 µm film thickness). Both columns were from Restek Corporation (Restek Corp., Bellefonte, PA, USA). The 2 D column was installed in a separate oven located inside the main GC oven. The system was equipped with a Gerstel MultiPurpose Sampler (MPS 2XL) with SPME option for procedural automation. The carrier gas was helium at a corrected constant flow rate of 2 mL min −1 and the injector operated in splitless mode. The main oven temperature program was 50 • C (5 min hold), a ramp of 5 • C min −1 to 300 • C (1 min hold). The secondary oven was programmed with a constant +10 • C offset relative to the primary oven. The modulation period was 2.0 s (0.3 s hot pulse and 0.7 s cold pulse time) with a +15 • C offset relative to the secondary oven. Mass spectra were acquired in the range m/z 40-500 at 200 spectra s −1 . The ion source temperature was set at 200 • C and the transfer line temperature was set at 250 • C. The detector voltage was run at an offset of -200 V relative to the tuning potential and the ionization electron energy (EI source) was set at 70 eV. Samples were acquired using LECO ChromaTOF software version 4.72.0.0 After smoking, filter pads were immediately placed in separate 20 mL headspace vials and sealed with magnetic crimp-top caps. Samples were stored in a refrigerator (≈4 • C) prior to analysis. The filter pads used for this chemical sampling were pre-fired in a muffle furnace at 750 • C for 4 hours. Cigarettes/joints, and filter pads were conditioned for at least 48 h at 60±3% relative air humidity and 22±1 • C prior to use 18 . Four different SPME fibre coatings were examined. The fibre coatings comprised a tri-mode (50/30 µm DVB/CAR/PDMS; divinylbenzene/carboxen on polydimethylsiloxane) fibre, a mixed-mode (65 µm PDMS/DVB; polydimethylsiloxane/divinylbenzene) fibre, a PDMS (100 µm polydimethylsiloxane) fibre, and a PA (85 µm polyacrylate) fibre, all purchased from Millipore Sigma (USA). All fibres were used to perform extractions from the headspace over the filter pads using the following conditions: an incubation time of 3 min at 70 • C and an extraction time of 15 min at 70 • C. The inlet temperature was maintained at 250 • C during fibre desorption (2 min). Fibres were initially conditioned according to the manufacturer's guidelines.
Data were processed using LECO ChromaTOF software version 4.71.0.0 with the following parameters. The expected peak width settings in the 1 D and 2 D were 12 s and 0.1 s, respectively. Peaks were detected from the raw chromatogram using a minimum signal-to-noise (S/N) ratio value of 100 with a minimum sub-peak S/N of 6. The minimum match required to combine peaks was 750. Identities of peaks were tentatively assigned on the basis of linear temperature-programmed retention indices (LTPRIs) for C5-C30 (LTPRI window ±10) and mass spectral similarity match (>750) against library spectra. Mass-spectral library searches were performed against the NIST/EPA/NIH Mass Spectral Library (NIST 17) and Wiley Registry of Mass Spectral Data (9th edition). Interactive LTPRI filters (±10) were performed by using the NIST/EPA/NIH Mass Spectral Library (NIST 17 version 2.3) as well as internet-based RI collections (i.e. PubChem) databases. Unless otherwise stated, all experiments were conducted with the aforementioned parameters. Detailed HS-SPME-GC×GC-TOFMS acquisition and data processing methods are provided at the end of the SI (SPME-GCxGC-TOFMS methods).

8/95
S4.2 SPME fibre selection Uptake of analytes in SPME is highly dependent on the chemistry of the fibre used to perform the extraction and in this study, four different SPME fibres were investigated for their utility in profiling volatiles: polydimethylsiloxane (PDMS), polyacrylate (PA), Divinylbenzene(DVB)/PDMS, and DVB/Carboxen(CAR)/PDMS. A comprehensive profile of the compounds in the smoke is desired, while ensuring good responses for compounds that are expected to be of interest, namely the terpenes and cannabinoids. During initial development of the GC×GC method, the PDMS fibre was abandoned as it gave poor recoveries for most compounds in tobacco and marijuana smoke. Typical chromatograms for both tobacco and marijuana smoke with the various fibre chemistries tested are included in Fig. S6.
Raw chromatograms consisted of several thousand peaks for each sample with the processing method employed. In curating the peak tables, unwanted peaks such as column bleed and extra hits from tailing peaks were removed. When we considered the total number of detected compounds, the PA fibre yielded 4329 and 1936 peaks for the particulate phase fraction of mainstream tobacco smoke and mainstream marijuana smoke, respectively. Similarly, using DVB/PDMS fibre 3680 peaks in the particulate phase fraction of mainstream tobacco smoke and 1845 peaks in the particulate phase fraction of mainstream marijuana smoke were detected. When the DVB/CAR/PDMS fibre was used, the total number of compounds detected increased to 4350 and 2575 for the particulate phase fraction of mainstream smoke from tobacco and marijuana, respectively. Although these values revealed the relative differences and complexity of the tobacco smoke and marijuana smoke, the DVB/CAR/PDMS fibre shows better extraction efficiency towards a larger number of analytes with diverse chemical functionality for both sample matrices. Hence this fibre type was selected for all the subsequent analyses.   -ND ND RDL is the reported detection limit of the analytical method (one-dimensional GC-MS or LC-UV) used by the contract laboratory performing quantitative analysis. Quantified compounds in two replicate samples employ the same units stated with RDL values. ND = not detected by GC×GC-TOFMS *Detected in tobacco only + Detected in marijuana only a Reported as "xylenes" at 25.5 mg/kg by external laboratory; individual quantification was not provided b Formaldehyde is a Group 1 carcinogen detected by external laboratory only. c Acetaldehyde is a Group 2B carcinogen (Group 1 carcinogen when associated with the consumption of alcoholic beverages) detected by external laboratory only. Table S3. Tentatively identified compounds from the particulate phase fraction of mainstream tobacco smoke. G1 = Group 1 carcinogen, G2A = Group 2A carcinogen, G2B = Group 2B carcinogen, G3 = Group 3 carcinogen, M = Mutagen, T = Teratogen, X = Toxic by other mechanisms, ND = No risk data available or risks mitigated by proper protective equipment.                                     This is the total flow into the inlet, which is the sum of the split flow, and column flow. It does not include the septum purge flow. When the total flow is changed, the split ratio and split flow change while the column flow and pressure remain the same.

3.5
Front Inlet temperature(s): No problems detected with oven temperatures.
Oven Equilibration Time ( sec ): 10 Note: All oven temperature ramps ( except the secondary oven ) will have the same duration. This is accomplished by extending the final hold time.
Enter oven temperature ramp below: The length of time from injection until the data system will start storing data from the mass spectrometer.

Check Retention Index Analytes
Maximum allowed retention index variation. 10 Update the retention times of retention index analytes.