J. Phys. Chem. A 2010, 114, 5967–5979
5967
High-Accuracy Measurements of OH Reaction Rate Constants and IR Absorption Spectra: CH2dCF-CF3 and trans-CHFdCH-CF3 Vladimir L. Orkin,* Larissa E. Martynova, and Alexander N. Ilichev† Physical and Chemical Properties DiVision, National Institute of Standards and Technology, Gaithersburg, Maryland 20899 ReceiVed: September 27, 2009; ReVised Manuscript ReceiVed: March 3, 2010
Rate constants for the gas phase reactions of OH radicals with two isomers of tetrafluoropropene, CH2dCF-CF3 (k1) and trans-CHFdCH-CF3 (k2); were measured using a flash photolysis resonance-fluorescence technique over the temperature range 220 to 370 K. The Arrhenius plots were found to exhibit a noticeable curvature. The temperature dependences of the rate constants are very weak and can be represented by the following expressions over the indicated temperature intervals:
k1(220-298 K) ) 1 . 145 × 10-12 × exp{-13 ⁄ T} cm3 molecule-1 s-1 k1(298-370 K) ) 4 . 06 × 10-13 × (T ⁄ 298)1.17 × exp{+296 ⁄ T} cm3 molecule-1 s-1 k2(220-370 K) ) 1 . 115 × 10-13 × (T ⁄ 298)2.03 × exp{+552 ⁄ T} cm3 molecule-1 s-1 The overall accuracy of the rate constant measurements is estimated to be ca. 2% to 2.5% at the 95% confidence level. The uncertainty of the measured reaction rate constants is discussed in detail. The atmospheric lifetimes due to reactions with tropospheric OH were estimated to be 12 and 19 days respectively under the assumption of a well mixed atmosphere. IR absorption cross-sections were measured for both compounds and their global warming potentials were estimated. 1. Introduction The internationally legislated elimination of chlorofluorocarbons (CFCs) from industrial applications, due to the danger that they pose to Earth’s ozone layer, has stimulated considerable study of the atmospheric properties of possible chemical substitutes. Chlorine-free partially fluorinated hydrocarbons (hydrofluorocarbons or HFCs) were among the leading environmentally acceptable CFC alternatives from the point of view of protection of the stratospheric ozone layer. They replaced CFCs in industrial applications with HFC-134a being the most widely known example. However, the rising concern about the possible impact of industrial chemicals on Earth’s climate has stimulated the further search of environmentally acceptable compounds. Quantification of the possible role of new compounds as “greenhouse gases” requires accurate information on their atmospheric lifetimes, which are key parameters in determining the environmental consequences following their release into the atmosphere. These data, when combined with IR absorption spectra, allow the estimations of radiative forcing and global warming potentials (GWP) through either thorough radiative-transfer modeling or semiempirical estimations. * To whom correspondence should be addressed. † Present address: Semenov Institute of Chemical Physics, Russian Academy of Sciences, Kosygin str. 4, 119991 Moscow.
The most important parameter in estimating the environmental impact due to emission of a compound is its residence time in the atmosphere (atmospheric lifetime), which is driven by the reaction of a compound with hydroxyl radicals (OH) for many atmospheric trace gases. The OH distribution and global average concentration can be derived from field measurements and budget calculations of a proxy atmospheric trace gas whose abundance is essentially controlled by reaction with OH and whose sources are well quantified.1-3 These studies suggest that the cumulative uncertainty in the OH concentration derived from such analyses ranges from 20 to 30%. For assessing the radiative impacts of a trace gas, the global warming potential (GWP)4 was adopted as a parameter analogous to the ozone depletion potential (ODP) introduced earlier for ozone depleting source gases.5,6 As in the case of ozone depletion, different atmospheric models can more accurately calculate the relative alterations in radiative forcing associated with two gases than the absolute climate response due to a change in the abundance of one gas alone. Indeed, the results of various modeling approaches for estimating radiative forcing and GWP are reasonably consistent. For example, radiative forcing calculations for HFC-134a suggested ca. 10% differences due to model differences.7 The calculated radiative forcing for other halogenated atmospheric gases usually agree to better than ca. 20% with the discrepancy of laboratory data partially responsible for this disagreement.8
10.1021/jp9092817 2010 American Chemical Society Published on Web 04/29/2010
5968
J. Phys. Chem. A, Vol. 114, No. 19, 2010
Orkin et al.
The sources of critically evaluated photochemical data for atmospheric modeling, NASA/JPL Publications9 and IUPAC Publications,10 recommend uncertainties within 10-60% for the majority of OH reaction rate constants with only a few cases where uncertainties lie at the low end of this range. These uncertainties can be somewhat conservative because evaluations are based on the data from various laboratories obtained during the last few decades. Nevertheless, even the authors of the original experimental works rarely estimate the total combined uncertainties of the published OH reaction rate constants to be less than ca. 10%. Thus, uncertainties in the photochemical properties of potential and current atmospheric trace gases obtained under controlled laboratory conditions still constitute a major source of uncertainty in estimating the compound’s environmental impact. One of the purposes of the present work was to illustrate the potential for obtaining accurate laboratory measurements of the OH reaction rate constant over the temperature range of atmospheric interest. Another source of uncertainty in GWP estimations is the accuracy of IR absorption measurements. In principle, such measurements should be easier to conduct than those of OH reaction rate constants. Nevertheless, IR measurement results from various laboratories often exhibit unexpectedly large differences. For example, in the case of the well studied IR absorption spectrum of HFC-134a differences among integrated band intensities obtained in various laboratories are as large as 10% for the strongest absorption features and increase to 25% for the weaker ones.7 The sources of such disagreements are not always obvious, but undoubtedly associated with systematic (instrumental) errors. Fluorinated alkenes are currently under consideration as potential environmentally friendly CFC substitutes. The presence of a carbon-carbon double bond is expected to render these substances highly reactive toward the hydroxyl radical resulting in an extremely short tropospheric lifetime, thereby limiting their accumulation in the atmosphere and minimizing their direct global warming potential (GWP). To quantify the atmospheric lifetimes and GWPs, we have investigated the reactivity toward OH of two isomers of tetrafluoropropene, 2,3,3,3-tetrafluoropropene and trans-1,3,3,3tetrafluoropropene:
OH + CH2dCFCF3 f products
(1)
OH + trans-CHFdCHCF3 f products
(2)
The quantitative results of their IR absorption cross sections measurements are also reported. The first reaction was originally studied in our laboratory a decade ago under complicated conditions when CH2dCFCF3 was found as an impurity in the sample of HFC-245cb (CH3CF2CF3).11 Later we studied a pure sample of the compound provided by Honeywell International Inc. Finally, we revisited this reaction by using our recently renewed flash photolysisresonance fluorescence apparatus to complete the study presented in this paper. This allows us to better estimate the actual instrumental uncertainties associated with our measurements of the OH reaction rate constants. The OH reactivity of the second compound, trans-CHFdCHCF3 was also investigated twice by using our apparatus before and after modification. Meanwhile, three papers devoted to the study of photochemical parameters
of these molecules have appeared,12-14 thereby allowing us to compare our results with those obtained by other research groups. Thus, the purpose of this paper is to provide accurate photochemical parameters for two compounds of potential industrial interest. At the same time, we are able to illustrate the possibility of determining OH reaction rate constants to an accuracy as small as 2-3%. In this work we also paid careful attention to both the IR measurement procedure and the gas handling to obtain accurate IR absorption cross sections. 2. Experimental Section15 2.1. OH Reaction Rate Constant Measurements. The general descriptions of the apparatus and the experimental method used to measure the OH reaction rate constants are given in previous papers.16-18 In the present work we used the same principal configuration of the apparatus with a number of modifications made to improve the accuracy of the obtained kinetic data by decreasing and quantifying the instrumental uncertainties. In particular, the gas handling system was completely replaced and a new reaction cell and photomultiplier were used. A brief description followed by the uncertainty analysis is given here. The principal component of the flash photolysis-resonance fluorescence apparatus is a Pyrex reactor (of approximately 180 cm3 internal volume) equipped with quartz windows. The reactor is thermostatted with methanol or water circulated through its outer jacket. This double-wall reactor is located in the metal vacuum housing evacuated to prevent ambient water condensation during low temperature measurements. It also prevents extraneous absorption of the UV radiation from the flash lamp used to produce OH radicals. Reactions were studied in argon carrier gas (99.9999 or 99.9995% purity) at a total pressure of 4.00-40.00 kPa (30.0-300.0 Torr). Flows of dry argon, argon bubbled through water thermostatted at 276 K, and fluoropropene mixtures (containing 0.1-100% of the reactant) flowed through the reactor at a total flow rate between 0.3 and 2.5 cm3 s-1, STP. The fluoropropene mixtures diluted with argon were premixed in glass bulbs (5 or 10 L) equipped with Teflon/glass valves (J. Young Scientific Glassware). The concentrations of the gases in the reactor were determined by measuring the gas flow rates and the total pressure with MKS Baratron manometers. Flow rates of argon, H2O/argon mixture, and the reactant/argon mixture were maintained, measured, and recorded using MKS mass flow controllers directly calibrated for every mixture. Hydroxyl radicals were produced by the pulsed photolysis (0.6-2.5 Hz repetition rate) of H2O, injected via the 276 K argon/water bubbler. The use of the below room temperature water bubbler assures a smaller and more stable concentration of water vapor in the reactor. The OH radicals were monitored by their resonance fluorescence near 308 nm, excited by a microwave-discharge resonance lamp (0.4 kPa or 3 Torr of a ca. 2% mixture of H2O in UHP helium) focused into the reactor center. Resonantly scattered radiation from the center of the reaction cell was collimated by the reactor window-lens and detected by a cooled photomultiplier working in the photon counting mode. The photomultiplier operation parameters were chosen so that photon counting obeys the normal Poisson statistics. The resonance fluorescence signal was recorded on a computer-based multichannel scaler (channel width 100 µs) as a summation of 500-5000 consecutive flashes. Therefore, the entire temporal OH profile was recorded and coadded following each flash, thereby minimizing any possible effect of small flash-
OH Reaction Rate Constants and IR Absorption Spectra
J. Phys. Chem. A, Vol. 114, No. 19, 2010 5969
to-flash variations in the initial OH concentration and drift of the detection radiation intensity. In the absence of the reactant (fluoropropene) in the reactor, the temporal decay of OH concentration is associated with the diffusion of OH out of the irradiation (photolysis) zone. This “background” diffusion-driven decay results in the noticeably nonexponential decay of the resonance fluorescence signal.17 In the beginning, the decay rate can be as large as ca. 90 s-1 at 30 Torr of Ar at room temperature; it can decrease down to ca. 20 s-1 at longer observation time of ca. 50 ms. This diffusiondriven decay rate depends on the pressure of Ar carrier gas in the reactor and its temperature. However, it remains very stable under the same conditions in the reactor. This background decay was usually recorded before and after a series of measurements at different reactant concentrations and was very stable under the normal operation conditions, usually better than 0.5-1 s-1 during multihour runs. (Actually, this background OH decay was very reproducible day-by-day and even month-by-month for the same H2O/Ar mixture, pressure, and temperature.) This is one of the advantages of using a nonreactive OH precursor, H2O. In the presence of the reactant under study, the OH decay rate increases accordingly.17 However, the recorded signal decay still does not follow a simple exponential decay, at least at smaller reactant concentrations. The procedure of correct deriving the reaction rate constant under such conditions of the diffusion-driven nonexponential background OH decay was described by Orkin et al.17 The procedure involves point-bypoint treatment of the data obtained in the presence of the reactant, I[TFP](t), and in its absence, I0(t), to calculate the decay coefficient due to the reaction of interest only while accounting for the actual shape of the “diffusional” temporal profile. Thus, the reaction rate constant kTFP can be obtained as
kTFP )
{
I0(t) 1 ∂ ln [TFP] ∂t I[TFP](t)
}
(3)
where [TFP] denotes the concentration of the reactant, tetrafluoropropene (TFP) under study. Such data treatment allows us to reliably measure reactive decay rates that are much slower than the background decay rate and yields pseudofirst-order plots with no intercept corresponding to the nonreactive background OH decay. Details of the technique can also be found in our subsequent paper.19 The fluoropropene concentration ranged from ca. 7 × 1012 to 6 × 1014 molecule/cm3 in our experiments. At each temperature the rate constant was determined from the slope of a plot of the decay rate versus the reactant concentration. The measured [OH] decay rate due to the reaction with fluoropropenes ranged from ca. 6 s-1 to ca. 430 s-1 in our experiments. The temperature points for the measurements were chosen to be approximately equally distant along the Arrhenius 1/T scale in order to have them equally weighted in the fitting procedure. Experiments were always performed at the two temperatures that are widely used in other studies, T ) 298 and 272 K. The first one is the standard (room) temperature used in the evaluations and presentations of the rate constants, and the second one is the temperature used in estimations of the atmospheric lifetime.1 All repetitive experiments were performed at the same standard temperatures. To check for any complications, test experiments were performed with the following variations of experimental parameters: the H2O concentration (a factor of 4), the flash energy (a factor of 4), the flash repetition rate (a factor of 4, between
0.6 and 2.5 Hz), the residence time of the mixture in the reactor (a factor of 8), the reactant concentration in the storage bulb (a factor of 100 for CH2dCF-CF3 and a factor of 330 for CHFdCH-CF3), and the total pressure in the reactor (between 30 and 300 Torr). No statistically significant changes of the measured reaction rate constant were observed in these experiments. Note, that variations of H2O concentration, the flash energy, and the flash repetition rate result in variations of the OH concentration in the mixture and, possibly, reactant photolysis and product accumulation. Thus, these test experiments were done to check for a possible influence of the secondary radical or product reactions on the measured rate constant. They allowed us to choose the appropriate “technical” conditions to avoid these potential chemical complications. The measurements performed with various gas flow rates and the reactant concentrations in the storage bulb allowed checking for the potential absorption or desorption of the reactant in the gas handling system and described below. Finally, the experiments at various pressures were done to check the possible pressure dependence of these reactions 1 and 2 between OH and unsaturated compounds. 2.2. IR Absorption Cross Sections Measurements. The IR absorption spectra were measured by using an FTIR spectrophotometer Nicolet 6700 with spectral resolutions of 0.125 cm-1 (recorded with a step of 0.06 cm-1) and 0.5 cm-1 (recorded with a step of 0.25 cm-1). Test experiments were also performed with spectral resolutions of 0.25 and 1 cm-1. All the IR absorption data presented in this paper were obtained by using a deuterated-triglycine-sulfate detector (DTGS) operated near room temperature. A liquid nitrogen cooled mercury-cadmiumtellurium detector (MCT) was also used for test measurements. Whereas the widely used MCT detector is much faster and more sensitive than DTGS, the latter one is characterized by better linearity and larger dynamic range. In addition, the spectral range of a DTGS detector extends to longer wavelengths. However, it is most important that the data obtained with an ambient temperature detector, DTGS, are free of those FTIR instrumental artifacts that affect the results of measurements with a cold MCT detector. These complications have been reported in the literature20-22 but have not received the proper attention of instrument makers and are often missed by users. We found that the problem becomes very pronounced at longer wavelengths when a low-temperature MCT detector is used to obtain the higher resolution quantitative absorption spectra. In this case, the small aperture’s iris located near the hot source of IR radiation (globar) gets warm and works like an additional noncollimated source of radiation for a cold detector. When a Fourier transform is performed, this results in a “phantom signal” that is especially pronounced at longer wavelengths (smaller wavenumbers). The presence of such an artifact can be easily seen when a FTIR apparatus shows the presence of radiation over the long wavelength region that is well beyond the optics transmission region (at wavenumbers smaller than 400 cm-1 in the case of KBr optics). When quantitative measurements are being performed this may result in the overestimation of the determined absorption cross sections if an MCT detector is used. We found this effect to be moderate for measurements with 0.5 cm-1 spectral resolution when a larger aperture can be used. It became very pronounced when data were collected with better spectral resolution of 0.125 cm-1. Therefore, we used only a DTGS detector and Boxcar apodization to obtain the data presented in this paper. The data were recorded with 32-128 scans. The photometric noise was as low as 0.0003 absorption units at larger wavenumbers (1000-2000
5970
J. Phys. Chem. A, Vol. 114, No. 19, 2010
Orkin et al.
cm-1) and increased up to ca. 0.003 absorption units at smaller wavenumbers (below 600 cm-1). A fast MCT detector was used only for preliminary measurements and in some test experiments to check the compound concentration in the storage bulbs. A (10.2 ( 0.05) cm glass absorption cell fitted with KBr windows was fixed in the spectrophotometer to minimize a baseline shift. The temperature of the cell was T ) (295 ( 1) K. Between measurements the cell was pumped out down to ca. 0.01 Pa and then filled with the gas to be studied. Absorption spectra of the evacuated cell and of the cell filled with a gas sample were alternately recorded several times, and the absorption cross sections at the wavenumber ν (cm-1) were calculated as
σ(ν) )
2.303 × (ATFP(ν) - A0(ν)) [TFP] × L
(4)
where σ(ν) is the absorption cross-section at wavenumber ν, in units of cm2 molecule-1; ATFP(ν) and A0(ν) are absorbancies (base 10) in the presence of tetrafluoropropene and of the evacuated cell at wavenumber ν, respectively; [TFP] is the concentration of tetrafluoropropene, in units of molecule/cm3; and L is the optical path length in cm. Measurements were performed at various pressures of a compound to verify adherence to the Beer-Lambert absorption law and obtain strong and weak absorption features of the spectrum. The final absorption spectra presented here were constructed from the results of individual measurements over the range of pressures where the measured absorption obeys the Beer-Lambert law. In our experiments the deviation from the Beer-Lambert law was not observed when the measured absorption, ATFP(ν), did not exceed ca. 0.9 for the majority of relatively smooth spectral features. Reproducibility of the measurements was better than 1% when the measured absorption exceeded ca. 0.1-0.2. Therefore, we conservatively considered the data derived from the measured absorptions over the range between 0.3 and 0.8 as the most reliable ones. The final absorption spectra presented here were constructed from the results of individual measurements of ATFP(ν) obtained between 0.3 and 0.8 only. The results obtained from these individual measurements were combined to exclude the data obtained at measured absorptions of either above 0.8 or below 0.3, except for the highest pressure data that were used to obtain the smallest absorption cross sections. Thus, the final spectra were constructed from the data points that obey the Beer-Lambert law and were obtained with reasonably large signal-to-noise ratios. The overall instrumental error associated with the optical path length, pressure measurements, temperature stability, and measured absorbance was estimated to be less than 2% for the strong absorption bands. 2.3. Instrumental Uncertainties. Since one of the purposes of this work was to demonstrate the possibility of higher accuracy of OH reaction rate constant measurements, we provide a detailed inventory of accountable sources of instrumental uncertainties related to our experiments. These estimated instrumental uncertainties should be combined with the uncertainty obtained from the statistical treatment of the results of the kinetic experiment. We attempted to eliminate all possible sources of systematic (type 2) uncertainties by making on-site calibrations of all instruments to estimate their accuracy. Thus, we obtained the statistical (type 1) uncertainties from the formal statistical treatment of calibration results instead. Pressure Measurements. One of two MKS Baratron manometers (100 Torr full scale or 1000 Torr full scale) connected to the reactor was used to measure and control the pressure in
the reactor. A manometer controls a downstream valve to maintain the reactor pressure to better than ca. (7 Pa (0.05 Torr). Two MKS manometers, 100 Torr (13.33 kPa) and 1000 Torr (133.3 kPa), were used to prepare the reactant mixtures. The majority of measurements were performed with a 1% mixture of CH2dCFCF3 and a 1.5% mixture of CHFdCHCF3 in Ar. In addition, 0.1, 0.3, 3, 10, and 100% mixtures were used for a number of test experiments. Two MKS manometers, 10 Torr (1.333 kPa) and 1000 Torr (133.3 kPa), were used to control the pressure in the absorption cell when IR absorption spectra were measured. All these manometers were adjusted and intercalibrated over the entire range of measurements by using the “laboratory pressure standard”, MKS Type 690A Baratron manometer with the stated standard accuracy of 0.08%. It was custom calibrated by MKS Instruments Inc. with the total stated uncertainty of less than 0.05%. In addition, this last manometer was directly calibrated using the NIST Primary Pressure Standard with the total uncertainty of less than 0.05% and was carefully handled after calibration. We conservatively assign the total relative uncertainty of pressure measurements to be 0.1% for our kinetic experiments. Thus, we have a small systematic uncertainty associated with this laboratory pressure standard. Temperature Measurements. The temperature in the reactor center (geometrical intersection of all three optical axes: the flash lamp, the resonance discharge lamp, and the photomultiplier) was calibrated by using 0.08 mm bare T-type (Copper/ Constantan) thermocouples. The temperature of incoming and outgoing liquid was measured along with the temperature in the reaction center and the temperature distribution in its vicinity at our standard temperatures (220, 230, 250, 272, 298, 330, and 370 K). In the actual kinetic experiments there was no thermocouple in the reactor center, and the temperature was determined from these calibrations based on measured temperatures of both incoming and outgoing liquid, which required a few tenths of a degree correction. Thermocouples were calibrated to be accurate to within 0.1 °C at 0 °C (melting ice) and 0.3 °C at 100 °C (boiling water). This uncertainty can gradually increase up to 1 °C at the lowest temperature T ) 220 K of our experiments.23 The spatial distribution of the temperature around the reactor center was also measured at all temperatures. The diameter of both the detection and photolysis zone in the reactor was less than 1 cm. The temperature deviation at ca. (0.5 cm off-center was ca. (0.1 K and (0.2 K at the highest (370 K) and lowest (220 K) temperatures, respectively. (It was less than (0.2 K and (0.4 K at 1 cm off-center at the same extreme temperatures). Temperature measurements introduce another source of systematic uncertainty that becomes increasingly important at low temperatures when no direct calibration of the thermocouple is available. Flow Rate Measurements. Gas flow rate measurements can be the major source of instrumental error in a kinetic experiment of higher accuracy. Therefore, we present the very detailed discussion of potential errors associated with gas flow measurements. All mass flow controllers were directly calibrated for every mixture by measuring the rate of pressure change in the same volume (the reaction cell isolated from the vacuum pump) by using the same reaction cell manometer. An additional volume could be connected to the reactor in case of larger flow rates of argon carrier gas. The ambient laboratory temperature could be slightly different day-by-day from ca. 22 to 24 °C, which can slightly affect the result of the calibration. The temperature of the cell was maintained equal to the ambient laboratory temperature ((0.2 °C) to have the entire calibration
OH Reaction Rate Constants and IR Absorption Spectra volume (the jacketed reactor with nonjacketed connecting tubes and the additional volume in case of Ar carrier gas calibration) at the same temperature. The temperature of all these parts of the vacuum system was measured by T-type thermocouples, as well as the temperature of the liquid circulating through the reactor jacket. The measured rates of the pressure change in the reactor were then normalized to T ) 298 K. The statistical uncertainty of such calibrations was usually within 0.1-0.2% for argon diluted mixtures, with any uncertainty larger than 0.2% being indicative of relatively poor quality of the particular calibration experiment. This uncertainty can be slightly larger for pure nondiluted reactants. There is an obvious reason for this larger uncertaintysthe actual gas flow rate of a pure compound is much smaller than for argon diluted mixtures at the same setting (set point) of a flow controller, which resulted in much longer measurements. Detailed analysis of potential errors associated with gas flow control and measurements is presented in the Appendix. Reactant Concentration in the Mixture. We used the reactants (fluoropropenes) premixed with argon in glass bulbs to feed a corresponding MKS flow controller. The mixtures were carefully prepared manometrically with attention paid to their thermal equilibration during the mixing. Note that the common procedure consisted of two parts: (i) careful filling a bulb with the reactant up to desired pressure of 0.1333-13.33 kPa (1.000-100.0 Torr) and (ii) dilution/pressurizing the reactant with high purity argon up to the total pressure of 133.3 kPa to obtain the desired mixture (0.1-10%). An accuracy of the first procedure is restricted by the accuracy of the calibrated manometer only. The pressurizing of the compound with a relatively large amount of argon unavoidably results in the warming of the gas mixture when argon is added to the bulb. Therefore, the total pressure in the bulb will settle down when both the bulb and gas mixture cool down to the room temperature. Hence, an additional amount of argon should be added upon thermal equilibration to have the desired mixture of known concentration not affected by this mixing warming. There are two obvious steps in this temperature equilibration process. First, warm argon gas thermally equilibrates across the bulb, which slightly warms the walls of the glass bulb, absorbing the heat. This temperature increase of the glass bulb was measured to be as high as up to ca. 1 K even at relatively slow flow of Ar to the storage bulb. Then, the bulb with the gas mixture slowly equilibrates to room temperature. “Complete” equilibration when both gas pressure and the temperature of the storage bulb stop decreasing can be as long as 20-40 min. This effect alone can increase the compound concentration in the mixture up to 0.3% if proper care is not taken. We controlled the bulb temperature and the pressure in the bulb to add argon until both pressure and temperature stop changing. A long mixture preparation procedure is needed to avoid this additional uncertainty. The most suitable concentrations to run flow controllers under the optimum conditions in our experiments were found to be 1% for CH2dCFCF3 and 1.5% for CHFdCHCF3. The accuracy of the mixture preparation depends on the precision of the manometers reading (0.1%) and the ambient temperature fluctuation (drift) during the mixing (less than 0.3 K). Thus, the uncertainty associated with the mixing procedure performed with the above-discussed precautions should not exceed ca. 0.2%. The most uncertain issue is a potential for reactant adsorption from the mixture (or desorption from the walls to the mixture) during storage in a bulb or when flowing through the gas handling system and the reaction cell. The reactant
J. Phys. Chem. A, Vol. 114, No. 19, 2010 5971 TABLE 1: Rate Constants Measured in the Present Work for the Reaction of OH with CH2dCF-CF3 (2,3,3,3-Tetrafluoropropene)a T (K) 220 230
250 272 298
313 330 350 370
k1(T) × 1012 (cm molecule-1 s-1)
[CH2dCFCF3] (1013 molecule/cm3)
1.079 ( 0.011 1.081 ( 0.008 1.108 ( 0.012 1.018 ( 0.024 1.013 ( 0.066 1.089 ( 0.011 1.091 ( 0.012 1.096 ( 0.007 1.103 ( 0.009 1.097 ( 0.022 1.087 ( 0.011 1.100 ( 0.009 1.100 ( 0.020 1.085 ( 0.021 1.109 ( 0.011 1.123 ( 0.016 1.140 ( 0.010 1.166 ( 0.012
1.6-16.7 2.4-31.6 2.7-16.7 2.7-22.6 1.7-21.3 1.4-16.2 2.0-13.5 1.8-24.4 2.0-20.6 2.0-17.1 2.3-19.3 1.2-12.1 2.0-16.3 2.2-27.7 1.7-23.3 1.7-15.5 1.6-20.8 1.5-19.9
3
test experiments conditions
13.3 kPa (100 Torr) 26.6 kPa (200 Torr) 40.0 kPa (300 Torr)
13.3 kPa (100 Torr) 26.6 kPa (200 Torr) total flow rate: 25% total flow rate: 200% 0.1% mixture in the bulb 10% mixture in the bulb
a The bold-highlighted data are results of the fit to all measurements performed at the particular temperature and 4.00 kPa (30.0 Torr) total pressure using a 1% mixture. These data are shown in Figure 1 and were used to derive the temperature dependences (5) and (6). The uncertainties are two Standard Errors from the least-squares fit of a straight line to the measured OH decay rates versus the reactant concentrations. They do not include any instrumental/systematic uncertainty.
concentration in a storage bulb was checked using IR absorption, and no measurable discrepancy between the content of the manometrically prepared mixtures and their IR analyses (accurate to better than ca. 0.5%) was found. The test kinetic measurements were performed at various total gas flow rates (a factor of 8) through the gas handling system and the reaction cell. Thus, we varied the residence time of the reactant in the gas handling system. Also, the test experiments were performed with very different reactant concentrations in the storage bulbsboth an order of magnitude larger and smaller than the optimal concentration used for the majority of measurements. There were no noticeable variations of the measured reaction rate constant obtained when various total flow rates and reactant mixtures were used, thus indicating the absence of complications resulting from compound wall adsorption/desorption in our measurements. Materials. The high purity samples of tetrafluoropropenes were provided by Honeywell International Inc. We studied two samples of trans-1,3,3,3-tetrafluoropropene (CHFdCHCF3) with 99.92 and 99.99% purity. The only detected impurity in both samples was 3,3,3-trifluoro-1-propyne (CHtCCF3) in the amounts of 0.08 and 0.01%, respectively. The sample of 2,3,3,3tetrafluoropropene (CH2dCFCF3) was ca. 99.9% purity with ca. 0.1% impurity of 3,3,3-trifluoropropene (CH2dCHCF3). These samples were used without any further purification except degassing. We used 99.9995 and 99.9999% purity argon (Spectra Gases Inc.) as a carrier gas. Note that the potential reactions between OH and reactive impurities in the samples pose another source of a systematic uncertainty in the reaction rate constant obtained from the experiment. 3. Results and Discussion 3.1. OH Reaction Rate Constants. The rate constants determined for the title reactions are presented in Tables 1 and 2. The bold-highlighted data are results of the fit to all measurements performed at the particular temperature and 4.00
5972
J. Phys. Chem. A, Vol. 114, No. 19, 2010
Orkin et al.
TABLE 2: Rate Constants Measured in the Present Work for the Reaction of OH with trans-CHFdCH-CF3 (trans-1,3,3,3-Tetrafluoropropene)a k2(T) × 1013 [CHFdCHCF3] T (K) (cm3 molecule-1 s-1) (1013 molecule/cm3) 220 230 240 250 260 272 285 298
313 330 350 370
7.39 ( 0.08 7.29 ( 0.06 7.19 ( 0.16 7.09 ( 0.07 7.11 ( 0.09 7.03 ( 0.09 7.05 ( 0.12 7.11 ( 0.05 6.89 ( 0.31 7.18 ( 0.14 7.30 ( 0.28 7.01 ( 0.27 7.19 ( 0.10 7.37 ( 0.11 7.49 ( 0.08 7.69 ( 0.07
2.6-30.0 2.5-33.0 2.4-18.4 4.5-30.6 4.3-25.2 2.1-39.8 4.0-38.0 2.0-60.0 3.5-22.7 3.5-34.8 3.9-34.2 14.0-43.0 3.6-34.6 3.4-32.9 3.2-31.0 3.1-29.3
test experiments conditions
26.6 kPa (200 Torr) 3% mixture 0.3% mixture 100% mixture
a
The bold-highlighted data are results of the fit to all measurements performed at the particular temperature and 4.00 kPa (30.0 Torr) total pressure using a 1.5% mixture. These data are shown in Figure 2 and were used to derive the temperature dependence (7). The uncertainties are two Standard Errors from the least-squares fit of a straight line to the measured OH decay rates versus the reactant concentrations. They do not include any instrumental/systematic uncertainty.
Figure 1. Results of the rate constant measurements for the reaction between OH and CH2dCFCF3, k1(T). b, this work; 0, Papadimitriou et al.;14 ], Nielsen et al.;12 O, Orkin et al.;11 ∇, our unpublished results obtained before the recent modifications. Uncertainties shown for our present data and data by Papadimitriou et al. are two standard errors from the fit whereas uncertainty reported by Nielsen et al. was chosen to encompass the extremes of their data. The dashed line is the best fit to our data with a three-parameter modified Arrhenius expression; the solid straight line is k1(220-298 K) ) 1.145 × 10-12 × exp{-13/T}, the solid curved line is k1(298-370 K) ) 4.06 × 10-13 × (T/298)1.17 × exp{+296/T}.
kPa (30.0 Torr) total pressure. These data were obtained with 1% (CH2dCFCF3) and 1.5% (CHFdCHCF3) mixtures prepared in the storage bulbs. These particular mixtures were chosen because they were the most convenient to maintain the appropriate reactant concentrations in the reaction cell with our apparatus design. The bold highlighted results are shown in Figures 1 and 2 along with other available data for these reactions. Arrhenius plots for these reactions exhibit curvature that is clearly resolved. Both rate constants have very weak
Figure 2. Results of the rate constant measurements for the reaction between OH and trans-CHFdCHCF3, k2(T). b, this work; ], Sondergaard et al.;13 ∇, our unpublished results obtained before the recent modifications. Uncertainties shown for our data are two standard errors from the fit whereas uncertainty reported by Sondergaard et al. was chosen to encompass the extremes of their data. The solid line is the best fit to our data with a three-parameter modified Arrhenius expression k2(T) ) 1.115 × 10-13 × (T/298)2.03 × exp{+552/T}.
temperature dependence; the variation of the rate constants over the entire temperature range is less than 10%. Nevertheless, our results clearly demonstrate these dependences with data scattering being quite negligible, thus illustrating the precision of measurements. Some results of test experiments are also shown in these tables. The test experiments were performed to check possible instrumental error that could increase the uncertainty of our results. The italicized results in the tables are indicative of the absence of any effect of gas flow rate (the residence time of the mixture in the system) on the measured rate constants. The variation of the reactant concentration in the storage bulbs by 2 orders of magnitude had no effect either. In addition, the quantitative IR analysis of the mixtures in the storage bulbs confirmed the content of manometrically prepared mixtures with the reactants concentrations as large as 10% and as small as 0.1%. We used the middle range concentrations of ca. 1% in our experiments. Thus, there was no indication of any additional systematic uncertainty in the reactant concentrations found other than those associated with pressure, temperature and flow rate measurements discussed above. Both reactions occur via addition of OH to an unsaturated C atom. Therefore, in spite of the multiatomic nature of the reactant molecules there is a possibility that these reactions are pressure dependent. The test experiments revealed no indication of any increase in the rate constants when pressure was increased by an order of magnitude. OH + CH2dCFCF3. A three-parameter modified Arrhenius dependence can be fit to the bold highlighted data set in Table 1 to result in the dependence shown in Figure 1 with the dashed line. In spite of a reasonably good fit to the individual points, one can see that this dependence does not reflect the trend at below room temperatures of atmospheric interest. Therefore, we suggest a simple Arrhenius dependence that was obtained from the fit to the data at T < 298 K and shown in Figure 1 with the straight solid line:
OH Reaction Rate Constants and IR Absorption Spectra
J. Phys. Chem. A, Vol. 114, No. 19, 2010 5973
k1(220-298 K) ) (1.145 ( 0.012) × 10-12 × -1 -1
exp{-(13 ( 3)/T} cm molecule 3
s
(5)
The uncertainties indicated here and everywhere in the text are two standard errors obtained from the statistical least-squares fit and do not include the estimated systematic uncertainty discussed earlier unless stated otherwise. Note that the variation of the reaction rate constant over the entire range of atmospheric interest is only ca. 1.5%. Thus, the average value is k1(220-298 K) ) (1.09 ( 0.02) × 10-12 cm3 molecule-1 s-1, with the error bars encompassing all the data obtained between 298 and 220 K with their statistical uncertainties. The above room temperature data can be represented by the following dependence shown in Figure 1 with the solid curved line for T > 298 K:
k1(298 - 370 K) ) 4.06 × 10-13 × (T/298)1.17 × exp{+296/T} cm3 molecule-1 s-1
(6)
The only detected impurity in the samples of CH2dCFCF3 was 3,3,3-trifluoropropene. The reaction between OH and CH2dCHCF3 is only about 50% faster than reaction 1 under study.11 Therefore, ca. 0.1% impurity of CH2dCHCF3 cannot affect the measured value of k1. There are published results from three studies available for this reaction. A decade ago we determined the rate constant of this reaction with no pure sample available:11 CH2dCFCF3 had been found as ca. 6% impurity in the sample of HFC-245cb (CH3CF2CF3). The disappearance of OH in that mixture of CH2dCFCF3 with CH3CF2CF3 was mainly due to reaction 1 because of 2 orders of magnitude difference in their reactivity toward OH.11 The main complication was the quantitative determination of the concentration of CH2dCFCF3 in that mixture. This was done by employing the photochemical titration of CH2dCFCF3 by molecular bromine followed by the quantitative photometric measurements of Br2 concentration.11 The results are shown in Figure 1 with open circles. These earlier data coincide with the results of our present most accurate measurements to better than 2%. There are two more recent publications on this reaction. Nielson et al.12 reported k1(296 K) ) (1.05 ( 0.17) × 10-12, cm3 molecule-1 s-1 as determined by a relative rate technique. This value coincides with our results within the reported uncertainty and is shown as a diamond in Figure 1. Papadimitriou et al.14 reported k1 obtained between 207 and 380 K, which are shown as squares in Figure 1. These authors reported the overall uncertainty of their data to be as large as 9% which overlaps the small difference with our results. All three available studies of k1(T) result in the same average temperature dependence of this rate constant over the common temperature range if one neglects the curvature of the Arrhenius plot. We can determine an average value below room temperature obtained by Papadimitriou et al.,14 k1(207-273 K) ) (1.095 ( 0.025) × 10-12 cm3 molecule-1 s-1, which encompass all the data with their statistical uncertainties. Thus, the average below room temperature value of k1 coincides with our average value shown above if one neglects the slow temperature dependence found in our study. Finally, triangles in Figure 1 show our unpublished results obtained for this reaction before the recent modifications. They exhibit the same temperature trend while being more scattered and slightly (ca. 2% in average) overestimated. The apparatus was modified to improve an accuracy of our data with great attention paid to gas handling
and all calibrations. Therefore, the bold highlighted data presented in Table 1 supersede all our previous results. OH + trans-CHFdCHCF3. The three-parameter modified Arrhenius dependence can be fit to the data set bold highlighted in Table 2:
k2(220 - 370 K) ) 1.115 × 10-13 × (T/298)2.03 × exp{+552/T} cm3 molecule-1 s-1
(7)
This dependence shown in Figure 2 well represents k2(T) over the entire temperature range. The detected reactive impurity in our trans-1,3,3,3-tetrafluoropropene samples was CHtCCF3. There are no data on OH reactivity of fluorinated propynes available. The analysis of the available data for the reaction between OH and nonsubstitute propyne,24 CHtCCH3, suggests that its rate constant does not exceed 1 × 10-11 cm3 molecule-1 s-1 over the entire temperature range of our study. Further, our purest sample contained only ca. 0.01% of CHtCCF3. Even if we accept the same rate constant for CHtCCF3, it cannot affect the results of our measurements of k2. Moreover, based on reactivity trends of fluorinated propenes, it is reasonable to suggest that fluorinated propyne is even less reactive. The only published study of this reaction was done by Sondergaard et al.,13 who reported the rate constant determined at 296 K, k2(296 K) ) (9.25 ( 1.72) × 10-12 cm3 molecule-1 s-1. In spite of the large reported uncertainty, this value significantly exceeds the result of our measurements at this temperature. The reason for this discrepancy is not clear. Both reactions 1 and 2 have similar rate constants. They were studied in the same research group, by using the same apparatus and the same relative rate technique with the same reference compounds, C2H4 and C2H2.12,13 Nevertheless, k1(296 K) is only 5% smaller than our value and coincides well within the reported uncertainty. However, the discrepancy in k2 is as high as 30% so that even large reported uncertainty does not extend it down to our value. Figure 2 also shows our unpublished results obtained for this reaction before the recent major modifications. They actually can be considered as independent measurements performed by using a flash photolysis-resonance fluorescence technique. They exhibit very similar temperature trend while being more scattered and slightly (ca. 2% in average) overestimated. Nevertheless, we believe that the accuracy of current measurements is noticeably improved and, therefore, the results presented in Table 2 supersede the previously obtained ones. Precision of OH Rate Constant Measurements. We intentionally increased a number of our standard temperature points where the rate constant k2(T) has been measured to doublecheck the precision of the present OH kinetic measurements. The statistical uncertainty usually used to present the results of such kinetic measurements can be misleading to some extent. The scattering of the original data points in the pseudofirst order plots (the OH decay rate versus the reactant concentration plots) does not obviously obey the normal statistical distribution. At the very least, it actually already includes some of the instrumental errors like drifts of flow controllers’ zero and temperature. Therefore, although the calculated standard errors from a least-squares fit are a measure of the scatter in the data, they probably should not be taken very strictly as the “true” confidence intervals of the normally distributed data set. This is rather the uniform and currently accepted way of data presentation and comparison. Therefore, the scattering of the finally derived average values of the rate constants around the
5974
J. Phys. Chem. A, Vol. 114, No. 19, 2010
Orkin et al.
TABLE 3: Summary of OH Reactivity of Fluorinated Propenes Containing -CF3
a
fluorinated propene
ki(298 K) × 1012 (cm3 molecule-1 s-1)
CH2dCH-CF311 CH2dCF-CF3 CHFdCH-CF3, transCHFdCF-CF3, cis-14 CHFdCF-CF3, trans-28 CF2dCF-CF311
1.5 1.1 0.7 1.3 2.2 2.2
E/R (T < 298 K) (K)a
E/R (298 K < T < 370 K) (K)a
-180 +13 -40 -170
-180 +90 +120 +20
-430
-410
From the fits to the data obtained in this work and those presented in refs 11 and 14.
smooth fitted line in the Arrhenius plot can serve as complementary evidence of the precision in addition to the reported formal statistical uncertainty of individual points. It would be the same value in the case of normal statistical distribution of the collected data. It provides an additional confidence in the “statistical” uncertainty we claim for our data. For example, the entire temperature dependence presented in Figure 2 spans 150 K along the temperature axis with ca. 9% variation of the measured rate constant, k2(T). In particular, the variation within eight points at below room temperature range of atmospheric interest is only 4.5%. Nevertheless, there is no scattering large enough to distort the obvious trend in the reactivitysthe best fit curve only emphasizes the already clear dependence. The deviation of the data points in Figure 2 from the best three-parameter fit (7) is in the range 0.1-0.7%, that is, less than one standard error for any of 12 data points shown in Figure 2. It is indicative that statistical uncertainties presented in Table 2 are overestimated. Total Instrumental Uncertainty and Accuracy of Measurements. Based on the above analyses we can estimate the total instrumental uncertainty associated with all the above-described accountable sources of uncertainty. The following is a summary of these sources. (1) Preparation of the reactant mixtures in the storage bulbs: two pressure measurements (0.1% each) and the temperature stability of the bulb (0.1%). (2) Pressure in the kinetic reactor: absolute pressure measurements (0.1%) and pressure fluctuations during the experiment (0.15%). (3) Flow controllers calibrations: Ar carrier gas flow (0.15%) and reactant mixture flow (0.15%). We can add the potential uncertainty due to fluctuation/drift of the flow controller “zero” as up to 0.1-0.2%. (4) Temperature of the reacting gas mixture in the photolysis/detection region: 0.05% at 298 K, 0.1% at 370 K, and up to 0.5% at 220 K. These uncertainties should be combined with statistical uncertainties associated with the results of kinetic measurements that are presented in Tables 1 and 2 and discussed above. The propagation of the experiment uncertainty is a controversial issue. We followed the formal approach and combined all statistically obtained uncertainties (Type 1) as square-root of the sum of their squares. Then systematic (Type 2) uncertainties were added to the result. These later uncertainties are associated with the calibration of the laboratory pressure standard manometer and the gas temperature measurements. We doubled these estimates to represent a 95% confidence level for estimated total uncertainty as ca. 2% at room temperature, which increases to ca. 2.5% at the lowest temperature of our experiments, 220 K. Our test experiments and detailed analysis did not reveal any other error that could be associated with either chemical complications in the reactor or the presence of the reactive impurities in the samples. Therefore, the above calculated total uncertainty should be considered as the estimation of the accuracy of our rate constant measurements. Discussion of Kinetics. The Arrhenius plots for both reactions exhibit weak but clearly resolved curvature. We do not expect
any simple mechanistic explanation of such behavior. The fits shown here were chosen to adequately present the measured temperature dependences only. Note that a curvature of the Arrhenius plot is not very uncommon when precise data are available.25-27 Only very general thoughts can be given on the nature of the curvature in case of abstraction reactions. We are afraid that its nature is obscured even more in case of OH addition reactions. Hopefully, such data can stimulate theoretical studies to improve our understanding of chemical mechanisms and developing predictive computational tools. We summarize in Table 3 the reactivity of fluorinated propenes containing the -CF3 group toward OH, where we show the room temperature rate constants, ki(298 K), and approximate temperature dependence, E/R, for above room temperature and below room temperature. Note that the CF3 substitution already dramatically decreased the reactivity of nonsubstituted propene by a factor of 20.29 The fluorination of olefinic carbons does not further change the reactivity appreciably. The changes in both absolute value of the OH reaction rate constant and its temperature trend are rather sporadicsthere is no obvious correlation between the substitution site and the corresponding reactivity change. The only valuable observation one can make is that the perfluorination of olefinic carbon atoms does not change the reactivity dramatically in comparison with nonfluorinated molecule (CF2dCF-CF3 vs CH2dCH-CF3). The same observation has been made11 in the case of ethene fluorination (CF2dCF2 vs CH2dCH2). In both cases, perfluorination of olefinic carbon atoms slightly increases the reactivity toward OH. 3.2. IR Absorption Spectra. IR absorption spectra of compounds under study are presented in Figures 3, 4 and 5. The measurements were done over the pressure range of 0.2-3.2 kPa (1.5-24 Torr) of CH2dCFCF3 and 0.27-8.5 kPa (2-64 Torr) of CHFdCH-CF3, respectively. The top panels in Figures 3 and 4 show the spectra obtained with a spectral resolution of 0.125 cm-1 (boxcar apodization) to illustrate the main absorption features. The lower panels show the same spectra recorded with a spectral resolution of 0.5 cm-1 in log absorption scale to illustrate smaller absorption features. IR absorption cross sections of both compounds are available in the Supporting Information. Note that these relatively large fluorinated molecules still exhibit a few sharp absorption peaks whose measured height depends on a spectral resolution and the bath gas pressure. This is rather pronounced in the case of CH2dCFCF3. Thus, its largest peak at 1181.5 cm-1 decreased by ca. 20% when recorded with the 0.5 cm-1 spectral resolution. The effect is even more pronounced for the small longer wavelength absorption band around 615 cm-1 illustrated in Figure 5. Nevertheless, a few sharp absorption features do not change substantially the integrated absorption and should not affect the estimations of global warming potentials. Infrared absorption spectra of these compounds were recently reported in refs 12, 13, and 14. Papadimitriou et al.14 reported integrated IR absorption band strengths of CH2dCFCF3 over a
OH Reaction Rate Constants and IR Absorption Spectra
Figure 3. IR absorption spectrum of CH2dCFCF3 obtained with a spectral resolution of 0.125 cm-1 (top panel) and 0.5 cm-1 (lower panel). The later is shown in log scale to visualize smaller absorption features.
J. Phys. Chem. A, Vol. 114, No. 19, 2010 5975
Figure 5. IR absorption band of CH2dCFCF3 between 560 and 640 cm-1 obtained with a spectral resolution of 0.125 cm-1 (top panel) and 0.5 cm-1 (lower panel).
CHFdCHCF3, whose peek absorption exceeds 1% of the maximum absorption of this compound at 1103.2 cm-1. The last column in Table 4 shows the relative difference with our results for comparison. The integrated intensity of the main absorption band of CH2dCFCF3 between 1065 and 1296 cm-1 reported by Papadimitriou et al.14 coincides with our results. However, there are noticeable differences for absorption bands around 950 cm-1, 1390 cm-1, and especially 1700 cm-1. The reason for these discrepancies is not clear. Both the DTGS detector and the cold MCT detector have relatively high sensitivity in this spectral range and the measured absorptions are large enough, at least, for the 1390 cm-1 band. In the meantime there is a surprisingly good agreement at the longest wavelength part of the spectrum where measurements can be complicated by both lower detectors sensitivities and the above mentioned FTIR instrumental artifact. The total absorption strengths integrated over the entire spectral range for both compounds obtained in our work are in agreement with those reported by Nielsen et al.12 and Sondergaard et al.13 within the reported uncertainties. 4. Atmospheric Implications
Figure 4. IR absorption spectrum of trans-CHFdCHCF3 obtained with a spectral resolution of 0.125 cm-1 (top panel) and 0.5 cm-1 (lower panel). The later is shown in log scale to visualize smaller absorption features.
number of spectral intervals while Nielsen et al.12 and Sondergaard et al.13 reported only the total IR absorption intensity integrated over the entire range of their measurements for both CH2dCFCF312 and trans-CHFdCHCF3.13 Table 4 presents all the available data on IR absorption intensity measurements. We accepted the wavelength intervals already chosen by Papadimitriou et al.14 for CH2dCFCF3. We calculated the integrated intensities of all resolved absorption bands of trans-
The atmospheric lifetimes of CH2dCFCF3 and transCHFdCHCF3 due to their reactions with tropospheric hydroxyl OH , can be estimated by using a simple scaling radicals, τTFP procedure that is based on the results of field measurements3 and thorough atmospheric modeling:1 OH τTFP )
kMCF(270) OH τ kTFP(270) MCF
(8)
OH where τOH TFP and τMCF are the lifetimes of tetrafluoropropene under study and methyl chloroform, respectively, due to reactions with hydroxyl radicals in the troposphere only; and kTFP(270 K) and kMCF(270 K) ) 5.9 × 10-15 cm3 molecule-1 s-1 (reference 9)
5976
J. Phys. Chem. A, Vol. 114, No. 19, 2010
Orkin et al.
TABLE 4: Integrated Infrared Absorption Band Strengths Obtained for CH2dCF-CF3 and trans-CHFdCH-CF3 (10-17 cm2 molecule-1 cm-1) integration range (cm-1) CH2dCF-CF3 540-655 850-920 920-1000 1065-1296 1307-1498 1580-1810 800-2000 CHFdCH-CF3, trans660-725 780-858 858-900 900-970 1045-1128 1128-1213 1213-1293 1293-1370 1650-1740 1740-1785 650-2000
this worka 0.483 0.635 0.375 12.04 2.94 0.783 16.90
Nielsen et al., 2007b
Søndergaard et al., 2007b
16.3 ( 0.9
0.436 0.137 0.353 0.726 5.05 5.93 2.01 2.19 2.31 0.17 19.59
19.4 ( 1.0
Papadimitriou et al., 2008c
deviation from this work, %
0.47 ( 0.05 0.63 ( 0.01 0.35 ( 0.01 12.0 ( 0.02 2.79 ( 0.02 0.66 ( 0.01
-2.8 -0.8 -7.1 -0.3 -5.4 -18.6 -3.7
-1.0
-1
Infrared absorption spectra were recorded at 0.5 cm spectral resolution and T ) 298 K. The overall total uncertainty is estimated to be ca. 2% (see the text). T ) 298 K. b Infrared absorption spectrum was recorded at 0.25 cm-1 spectral resolution and T ) 296 K. Reported uncertainty includes the estimated systematic error c Infrared absorption spectrum was recorded at 0.5 cm-1 spectral resolution and T ) 296 K. The quoted uncertainties are two standard error from the statistical treatment of the data only. a
are the rate constants for the reactions of OH with these OH ) 6.0 years was substances at T ) 270 K. The value of τMCF obtained from the measured lifetime of MCF of 4.9 years when an ocean loss of 89 years and a stratospheric loss of 39 years were taken into account. Applying this method to the title compounds of this study yields the estimated atmospheric lifetimes of 12 and 19 days for 2,3,3,3-tetrafluoropropene and trans-1,3,3,3-tetrafluoropropene, respectively. The lifetimes derived here are very short, much shorter than the characteristic time of mixing processes in the troposphere and hence are only crude estimates. The use of eq 8 is applicable for long-lived species that are well mixed throughout the troposphere. Its use for short-lived atmospheric gases having lifetimes shorter than the characteristic time of mixing processes in the troposphere can result in significant errors due to the large spatial gradients of their concentration in the atmosphere. For such species, eq 8 provides only rough estimates of the tropospheric lifetimes with respect to reaction with OH. The correct residence time of short-lived compounds in the atmosphere will depend on the emission location and season as well as local atmospheric conditions.30,31 Nevertheless, the results of these modeling studies demonstrate that such an estimation procedure gives reasonable average values30,31 and provides a useful scaling of the lifetimes of such compounds. These very reactive unsaturated compounds can also react with other active components of the troposphere, Cl, O3, NO3. The rate constants for reactions of both compounds with Cl and O3 were determined by Nielsen et al.12 and Sondergaard et al.13 at T ) 296 K. To the best of our knowledge, there are no data available for their reactions with NO3. The possible effect of all these reactions on the residence time of the compounds in the atmosphere was already discussed in refs 12, 13, and 14. It is very clear that their reactions with tropospheric ozone are too slow to change the atmospheric lifetime. Based on the O3 reaction rate constant k(TFP + O3) ) 2.8 × 10-21 cm3 molecule-1 s-1 reported for both compounds,12,13 their globally averaged atmospheric sink due to the reaction with O3 is
characterized by ca. 13 years atmospheric lifetime. Therefore, even in the polluted industrial areas with larger ozone concentrations this atmospheric sink is not comparable with OH reactions. The situation is very different with Cl and NO3 reactions. These reactions can significantly shorten the atmospheric lifetime of the compounds under study especially in some locations and periods of time. Unfortunately, in great contrast with OH, these changes are much more difficult to quantify presently. The rate constants of the reactions between tetrafluoropropenes under study and Cl are much larger than those with OH.12,13 On the other hand, the average concentration of Cl atoms in the troposphere is much smaller than that of OH radicals, although large fluctuations are expected in some regions of the marine boundary layer.32 Some estimations allow it to be as large as 104 molecule/cm3 in the free troposphere.32,33 Assuming such a globally average concentration of chlorine atoms we can estimate the atmospheric lifetimes due to reactions with Cl only to be as short as 16 and 25 days for CH2dCFCF3 and trans-CHFdCHCF3, respectively. These lifetimes are roughly the same as OH reaction related lifetimes estimated above. If we accept these Cl reaction lifetimes, the global atmospheric lifetimes of CH2dCFCF3 and trans-CHFdCHCF3 will become as short as 7 and 11 days, respectively. Again, these estimations are based on the assumption of the global average concentration of atmospheric Cl as large as 104 molecule/cm3, which is not established well enough to make definite conclusions. It is an even worse situation for estimating the possible effect of NO3 reactions as the atmospheric sink because neither NO3 reaction rate constants nor its atmospheric concentration are known. The available rate constants at T ) 298 K for reactions of NO3 with analogous molecules are 9.5, 3.8,
