Anal. Chem. 2002, 74, 5871-5881
Evaluation of Hydrazine Reduction by Cellulose Acetate Filters Using Infrared Tunable Diode Laser Spectroscopy Charles N. Harward,† Milton E. Parrish,* Susan E. Plunkett, Joseph L. Banyasz, and Kenneth H. Shafer
Nottoway Scientific Consulting Corporation, P.O. Box 125, Nottoway, Virginia 23955, and Philip Morris USA Research Center, 4201 Commerce Road, Richmond, Virginia 23234
Cellulose acetate (CA) filters have been investigated to determine their hydrazine (N2H4) breakthrough characteristics using a system based on tunable diode laser absorption spectroscopy (TDLAS). The breakthrough mass loading sorption curves for hydrazine were dependent on both the flow rate and the concentration. In experiments using a 4.5 ppmv hydrazine standard, the amounts of hydrazine retained by the CA filter were 4.25 µg at a flow rate of 2.82 L/min and 65 µg at a flow rate of 0.28 L/min. These loadings are much greater than the 31.5 ng/cigarette of hydrazine reported in smoke for unfiltered cigarettes. Further, CA filters exposed to four and eight puffs of smoke actually made the filter more efficient in retaining hydrazine compared to CA filters that had not been exposed to smoke. Therefore, if hydrazine is present in smoke at the levels reported in unfiltered cigarettes, all of the hydrazine would be trapped by the CA filter, and would be unable to break through during smoking. A unique feature of this analytical method is that the instrument does not require calibration after molecular parameters have been determined, in this case from previously acquired quantitative hydrazine FT-IR reference spectra. Hydrazine (N2H4) was reported to be present in tobacco smoke collected from 20 unfiltered domestic commercial cigarettes using a very involved multistep process.1 The approach required reacting the pentafluorobenzaldehyde complex with any hydrazine that may be in the smoke to form the decafluorobenzaldehyde azine (DFBA) derivative. The DFBA derivative was isolated from the smoke matrix using thin-layer chromatography with two different plate media and developing solvents, followed by separation and detection using gas chromatography with electron capture detection. The need for such an extensive analytical method for measuring hydrazine in tobacco smoke is due to the complexity of the smoke sample matrix, the very low concentration, and the high reactivity of hydrazine. This method did not offer the necessary sensitivity for analyzing the fresh smoke of individual puffs. The per-puff measurement approach minimizes aging effects * To whom correspondence should be addressed. Phone: (804) 274-3490. Fax: (804) 274-2886. E-mail:
[email protected]. † Nottoway Scientific Consulting Corporation. (1) Liu, Y.; Schmeltz, I.; Hoffmann, D. Anal. Chem. 1974, 46, 885-889. 10.1021/ac0257187 CCC: $22.00 Published on Web 10/24/2002
© 2002 American Chemical Society
that result from having to collect the entire sample over 8 to 10 min of smoking. A system based on tunable diode laser absorption spectroscopy (TDLAS) was used to attempt to detect hydrazine in single puffs of fresh tobacco smoke.2-5 The sampling system was designed to inject single puffs of whole smoke of 2-s duration directly from the cigarette in a flowing stream of nitrogen into the infrared gas cell. The per-puff measurement was completed in 3 s. Although the limit of detection (LOD) for hydrazine in nitrogen in the gas cell was determined to be 0.09 ppmv, the inability to adequately address the absorption of unidentified smoke components resulted in an LOD of 4.2 ppmv for a single puff of smoke. Consequently, we could not confirm whether hydrazine was generated in a single puff of fresh smoke. However, the speed and selectivity of the instrument was sufficient to demonstrate that 83% of a 520 ppmv hydrazine standard added to the smoke matrix was consumed in the first second of a 3-s puff sample.3 This observation was explored further by studying the gas-phase kinetic mechanism of the hydrazine and acetaldehyde reaction (the major aldehyde present in tobacco smoke). It was found that the rate of reaction was enhanced by a factor of 100 in the presence of cellulose acetate (CA) filter material. This is significant because most cigarettes today have CA filters. Manufacturers first used such filter material beginning in the early 1950s. CA filtered cigarettes represented 65% of the U.S. market by 1965 and 96% by 1985.6,7 By 1998, the market share had risen further to 98%.8 Since the only published report detecting hydrazine in tobacco smoke was based on unfiltered cigarettes obtained in 19721, we thought it was important to determine the retention and breakthrough behavior of hydrazine for typical CA filters using similar sampling conditions described for cigarette (2) Plunkett, S.; Parrish, M.; Shafer, K.; Nelson, D.; McManus, J. B.; Jimenez, J. L.; Zahniser, M. SPIE Proc. 1999, 3758, 212-220. (3) Plunkett, S.; Parrish, M.; Shafer, K.; Nelson, D.; Shorter, J.; Zahniser, M. Vibr. Spectrosc. 2001, 27, 53-63. (4) Plunkett, S.; Zahniser, M.; Parrish, M.; Nelson, D.; Shorter, J.; Wormhoudt, J.; Shafer, K. J. Mol. Spec. 2002, 211, 241-247. (5) Plunkett, S.; Parrish, M.; Shafer, K.; Shorter, J.; Nelson, D.; Zahniser, M. Spectrochim. Acta, Part A 2002, 58, 11, 2505-2517. (6) Kiefer, J. E.; Touey, G. P. In Tobacco and Tobacco Smoke; Wynder, E. L., Hoffmann, D., Eds.; Academic Press: New York, 1967; p 566. (7) Browne, C. L. The Design of Cigarettes, 2nd ed.; Celanese Fibers Marketing Company: Charlotte, NC, 1981; p 40. (8) Stratton, K.; Shetty, P.; Wallace, R.; Bondurant, S., Eds. Clearing the Smoke: Assessing the Science Base for Tobacco Harm Reduction. National Academy Press: Washington, DC, 2001, p 26.
Analytical Chemistry, Vol. 74, No. 22, November 15, 2002 5871
Figure 1. This schematic representation shows the inlet sampling system for the evaluation of CA filters for reduction of hydrazine using TDLAS.
smoke analysis reported previously.3 This paper describes the experimental design, the sampling procedure, and the results from experiments in which the effects due to changes in flow rates and hydrazine concentrations are studied. For a 4.5 ppmv hydrazine standard at a flow rate of 2.8 L/min, 4.25 µg was retained by the CA filter. This means that the CA filter retains at least 135 times the 31.5 ng/cigarette hydrazine delivery reported to be present in smoke for unfiltered cigarettes.1 A unique feature of this analytical method is that the instrument did not require calibration with known standards after the molecular parameters had been determined, in this case from previously acquired quantitative hydrazine FT-IR reference spectra. EXPERIMENTAL SECTION Instrumentation. The hydrazine concentrations were monitored using a system with dual lead salt infrared tunable diode lasers. It is a modified version of the system described in a previous publication.2 A liquid nitrogen dewar (MN ND-5, Infrared Laboratories) housed lead salt tunable diode lasers (Laser Components Inc.) and HgCdTe detectors (Kolmar Technologies). For this work, only one of the two diodes was used. The diode was operated with >93% of the energy in a single mode as determined by adding sufficient ammonia (NH3) to the sample cell to obtain complete absorption of the dominant mode. The energy from the diode laser was collected and collimated by all reflective optics and then split into two beams. One of the beams was passed through a reference gas cell containing pure ammonia and was used for monitoring the laser wavelength. Periodically, the laser wavelength was relocked to an ammonia absorption line. This locking procedure using ammonia was necessary to compensate for system drifts, especially for the hydrazine experiments requiring hours to complete. The other beam was directed through a multipass astigmatic Herriot cell set to a path length of 100 m (model AMAC-100, Aerodyne Research Inc.). Front surface mirrors directed the beam through the cell to one of the detectors, where it was converted into an analogue signal. Gas Sampling System. A continuous flow inlet system minimized the loss of hydrazine due to sorption on or reaction 5872
Analytical Chemistry, Vol. 74, No. 22, November 15, 2002
with the cell walls and associated tubing. Figure 1 shows a schematic representation of the sampling system for the 100-m multipass cell. Due to the nature of hydrazine, it is important to verify that the exhaust hood is operating properly. Nitrogen flowing through a cooled bubbler containing pure liquid hydrazine (99% anhydrous, 2-mL ampules, Aldrich) was used to generate a range of hydrazine standards. The ampules were stored in a refrigerator until needed. Reasonable caution must be used when handling liquid hydrazine to fill the bubbler. Wearing appropriate gloves for protecting the hands was required. This procedure must be performed in a properly maintained exhaust hood, since hydrazine is a strong reducing agent and can cause injury to skin and eyes. Lowering the temperature of the hydrazine in the bubbler reduced the vapor pressure and the rate of consumption of hydrazine. A Haake model DC 10 stirrer bath controlled the temperature of the bubbler and a Thermo Neslabs model CFT25 refrigerated recirculator provided the cooling for the bath. A computerized gas blending system (Environics Inc., model S-4000) controlled the flow rate of the nitrogen through the bubbler. The flow rate could be varied from 3 to 130 mL/min. The dilution flow was controlled by another computerized gas blending system (Environics, Inc., model 2000). This Environics unit also was used to blend in ammonia with the hydrazine and nitrogen. The ammonia, which is a hydrazine decomposition product, was used to determine the response of the system as well as the response of CA filters to it. Additional flow controllers from MKS (model 1259C) were used for dilution flows of nitrogen up to 50 L/min. The total flow was always maintained higher than the flow rate into the 100-m cell. The excess gas sample flowed into a negative pressure hood that was vented to the outside atmosphere. A rotary vacuum pump at the cell outlet, along with a critical flow orifice, maintained a constant flow rate into the cell. A throttle valve at the exit of the 100-m cell was required to maintain the cell pressure at 24 Torr with a flow rate of 1.27 L/min. This pressure provided an optimal compromise between narrowness and depth of the molecular absorption lines. This flow rate is slightly above the 1.05 L/min required to generate a conven-
tional 35-mL puff volume in a 2-s puff duration during cigarette smoke analysis. The manual valves in the inlet system were used to control the flow of hydrazine or nitrogen. By the use of the appropriate on/off valves, hydrazine or nitrogen (used for clearing the sampling system) could flow through the filter compartment or the bypass port. CA filters could be placed in or retrieved from the filter compartment when the flow was directed through the bypass port. The bypass assembly was necessary to minimize the amount of room air entering the sampling system. The room air would disrupt the equilibrium concentration of the gases being studied. The moisture in the air would cause both hydrazine and ammonia to be displaced from the inlet and cell walls. In addition, this bypass assembly reduced changes in cell pressure that would otherwise disrupt the hydrazine/ammonia equilibrium. Atmospheric water and carbon dioxide were removed from the nitrogen supply used in these experiments using a Whatman FT-IR purge gas generator model 75-62. The CA filters used in the study were removed from Philip Morris monitor cigarettes, the specifications of which have been previously reported.9 Additional specifications for the CA filter are 0% ventilation, 20.8 mm length, 24.5 mm circumference, 68 mm of water (5 Torr) resistance-to-draw (RTD) pressure drop, glyceryl triacetate (triacetin, C9H14O6) applied as plasticizer at 8 wt %, cellulose acetate fiber denier per filament (dpf) of 2.7 for total denier of 35 000, and plug wrap with lap and anchor adhesives. A brief description of cellulose acetate filter tow technology, the manufacturing process, and how the CA filter is used with cigarettes for filtration is discussed by several researchers (including references within).7,10 Spectral Data Acquisition. Molecules are detected by observing their absorption features in a spectrum generated by periodically sweeping the current through a lead salt diode laser and synchronously monitoring its intensity with an infrared detector. The operating temperature of the diode and the average current through the diode controlled the central wavelength (966.40 cm-1), whereas the fine wavelength tuning was accomplished by ramping the current applied to the diode. Spectral data were collected via a 12-bit 3.4 MHz analog-to-digital converter from National Instruments at a digitization rate of 500 kHz. Each spectrum consisted of 400 points taken as the laser current was scanned through a 16.6 mA ramp. Twenty points at the end of the ramp were used to drop the laser current below threshold in order to determine the zero light levels on the AC-coupled detectors. The spectral data used in this study were the result of co-adding 6250 individual spectra for a 5-s collection time. The spectrum was fit using a polynomial baseline, the nonlinear tuning rate of the laser, the zero light level, the temperature and pressure in the 100-m cell, and selected hydrazine and ammonia line positions to determine their molecular concentrations. The molecular spectral parameters for ammonia were obtained from the high-resolution transmission molecular absorption (HITRAN) database.11 The hydrazine line parameters were determined by deconvolving high-resolution hydrazine FT-IR spectra. A macro program written for the IGOR (Wavemetrics, Inc., Lake Oswego, OR) data analysis software package performed the (9) Parrish, M. E.; Harward, C. N. Appl. Spectrosc. 2000, 54, 1665-1677. (10) Norman, A. In Tobacco Production, Chemistry and Technology; Davis, D. L., Nielson, M. T., Eds.; Blackwell Science: London, 1999; pp 353-387.
deconvolution. Complex spectra for compounds such as hydrazine necessitate the use of this procedure as a result of the blending of the Doppler broadened lines. The macro, SELECTLINES, allows the operator to select the relative line positions from the absorption spectrum, and then it estimates the linestrengths from the absorption spectrum. It calculates a blended line absorbance spectrum from the selected individual absorption lines by convolving the relative line positions and linestrengths with a Voigt line shape and the assumed instrumental broadening function. The broadening coefficient for the absorption lines had been previously determined.4 Lines are added, and the strengths are adjusted until a best match with the FT-IR spectrum is obtained. This procedure does not lead to a unique solution for line positions and linestrengths, since the number of lines and their strengths can be arbitrarily adjusted. After the molecular parameters have been determined, the TDLAS system does not require periodic recalibrations, since the molecular parameters are used in the fitting procedure. We have shown previously in our laboratory that the fitting procedure provided quantitation that is within stated accuracies of standards for ammonia when using the ammonia and ethylene molecular parameters from the HITRAN database.3 The accuracy of the molecular parameters determined from the deconvolution determine the uncertainty in the hydrazine. The overall accuracy for hydrazine, the molecular parameters, and the quantitation of hydrazine are estimated to be within (7%.4 The use of this quantitation method is important for gases such as hydrazine that cannot be made into reliable gas mixtures because of their reactivity or tendency to polymerize. It removes the need to use bubblers or gas permeation devices once the molecular parameters have been determined. Sampling Procedure. All of the breakthrough experiments were performed in a conditioned laboratory (60 ( 1% RH, 22.5 ( 0.5 °C) using the following procedure: The system baseline signal was determined with flowing nitrogen with no filter in the sample compartment (see Figure 1). The response of the system to hydrazine and ammonia (hydrazine decomposition product) was determined by flowing a standard of hydrazine in nitrogen through the filter compartment containing no filter. This hydrazine standard was allowed to flow through the system into the gas cell until the response for hydrazine reached the maximum level, which was referred to as the steady state condition. At this point, the hydrazine standard was turned off, nitrogen continued to flow through the sampling inlet and the gas cell, and the hydrazine response returned to baseline. The system response to hydrazine and ammonia was completed after the baseline was reached. Next, the response of the CA filter was determined. At this point, the bypass valves were opened, the valves before and after the filter compartment were closed, and a fresh CA filter was inserted into the filter compartment. After the filter was inserted and the nitrogen flow was returned to the filter compartment, the system baseline was perturbed slightly because of atmospheric moisture entering the filter compartment when the CA filter was inserted. This small amount of moisture released hydrazine and ammonia from the cell walls as it passed through the system. The system (11) Rothman, L. S.; Rinsland, C. P.; Goldman, A.; Massie, S. T.; Edwards, D. P.; Flaud, J.-M.; Perrin, A.; Camy-Peyret, C.; Dana, V.; Mandin, J.-Y.; Schroeder, J.; McCann, A.; Gamache, R. R.; Wattson, R. B.; Yoshino, K.; Chance, K. V.; Jucks, K. W.; Brown, L. R.; Nemtchinov, V.; Varansasi, P. J. J. Quantum Spectrosc. Radiat. Transfer 1998, 60, 665-710.
Analytical Chemistry, Vol. 74, No. 22, November 15, 2002
5873
Figure 2. The top spectrum is ammonia (NH3) at 80.7 ppmv, and the bottom spectrum is hydrazine (N2H4) at 40 ppmv for the 100-m cell at a pressure of 11.4 Torr. The axes on the right-hand side of the spectra are for the line strengths of the two molecules. The vertical lines in the figure represent the line strengths in units of cm2 molecule-1 cm-1. The tuning rate of the TDL was highly nonlinear in the first 100 data points, and these data are not included in the figure.
was allowed to return to a steady-state level before the hydrazine standard was turned on again. After the hydrazine standard in the presence of the CA filter reached a steady-state condition, the hydrazine flow was turned off, and the system response returned to baseline. In this paper, experiments with the CA filter in the system and a flowing hydrazine standard are referred to as CA filter breakthrough or sorption curves. The data taken after the hydrazine standard is turned off are referred to as CA filter desorption curves. Experiments were performed using the above procedure to evaluate the CA filter for the retention of hydrazine at different concentrations and flow rates as well as for filters previously exposed to cigarette smoke. In addition, the desorption of CA filters previously saturated with hydrazine was studied. RESULTS AND DISCUSSION Characterization of the TDLAS Instrument and Sampling System. The objective of this work was the determination of the adsorption/absorption of different initial concentrations of hydrazine in nitrogen standards on CA filters in order to determine the breakthrough characteristics of the CA filter. There is no distinction made in this study between adsorption and absorption, because the technique used cannot distinguish between the two types of phenomena. Therefore, in this paper, we refer to either term as sorption. Since hydrazine decomposes to ammonia, nitrogen, and hydrogen,12 and ammonia absorbs in the same infrared region as hydrazine, one must know the spectral features of each species at the wavelengths used for quantitation. Figure 2 shows the spectra for hydrazine and ammonia taken with the (12) Schmidt, E. W. Hydrazine and Its Derivatives, Preparation, Properties, Applications; John Wiley and Sons: New York, 2001; Vol. 1, pp 1105-1111.
5874 Analytical Chemistry, Vol. 74, No. 22, November 15, 2002
TDLAS system for the spectral region 966.32-966.55 cm-1. The hydrazine standard was 40 ppmv, and the ammonia concentration was 81 ppmv at a cell pressure of 11.4 Torr. The figure shows that the maximum absorption line strength of ammonia was much higher (∼27 times) than for hydrazine. The region between 966.400 and 966.545 cm-1 was used for the spectral fitting to determine molecular concentrations. Also shown in this figure is the least-squares fit to the measured spectrum. The two hydrazine spectra are almost identical and therefore have been offset by 0.01 absorption units in order to distinguish between the two spectra. The strong ammonia and hydrazine features to the left of the fitting region were not used because they are overlapping and are in a region where the diode tuning characteristics are nonlinear. Use of this region in the fitting process makes the standard deviations of the fit higher. The absorption line at 966.4736 cm-1 for ammonia was used in the reference beam to lock the laser wavelength. The multiple vertical lines in the hydrazine spectra in the figure represent the integrated line strengths determined from the deconvolution of the FT-IR spectra. The value for the largest hydrazine integrated line strength from the spectra is a factor of 1.79 higher than the largest value we reported previously for a different spectral region (965.4-965.6 cm-1).4 The higher integrated strength for the hydrazine improved the system sensitivity. Another attribute of this analytical instrument system is the capability to monitor the ammonia decomposition product of hydrazine. This helps in minimizing any potential adverse effects on the least squares spectral fit, which directly influences the quantitation of hydrazine. Figure 3 shows the results for a typical experiment at a 10 ppmv hydrazine concentration with a flow rate of 1.27 L/min. The ammonia in the top trace is the result of decomposition of the
Figure 3. The lower curve shows the system response at 1.27 L/min of a 10 ppmv hydrazine (N2H4) standard with and without the CA filter. The upper curve shows the ammonia (NH3) residual at ∼0.1 ppmv from hydrazine decomposition with and without the CA filter. The disturbance in the baseline in both curves at ∼1 h 40 min occur when the filter is placed in the sample compartment. Table 1. Breakthrough Time and Time to Reach 50% Steady State with and without CA Filter for a 10 ppmv Hydrazine Standard and Ammonia Residual from Hydrazine Decomposition no filter, hydrazine @10 ppmv (s)
filter, hydrazine @10 ppmv (s)
time to
for hydrazine
for ammonia
for hydrazine
for ammonia
breakthrough 50% steady state
