CF₃SF₅ - a 'super' greenhouse gas

Over the past seven years the importance of greenhouse gases and global warming has leapt from obscurity to the top of the political agenda in all developed countries, culminating in the Stern report in November 2006 when the economic consequences of unchecked global warming were spelt out. Here we review the science of the greenhouse effect - or, more accurately, radiation trapping - and describe what constitutes a serious greenhouse gas, taking CF₃SF₅ as a case study. 

CF₃SF₅  in the atmosphere 

CF₃SF₅ is a gas at room temperature with a boiling point of 253 K and an enthalpy of vaporisation of 20 kJ mol^-1. The gas was first detected in the Earth's atmosphere in 2000,¹ by which time the related SF₆ molecule had already been identified as a potential greenhouse gas.² The source of CF₃SF₅ is believed to be anthropogenic, and most likely a breakdown product of the dielectric molecule SF₆  in high-voltage equipment. The trends in concentration levels of SF₆ and CF₃SF₅ have tracked each other closely over the past 30-40 years (Fig 1 (a)), suggesting that CF₃SF₅ is probably produced in the electronics industry via the recombination of CF₃ and SF₅ free radicals.  

Moreover, measurements taken from ice samples in Antarctica show similar variations of concentration of SF₆ and CF₃SF₅  at depths below the Earth's surface (Fig 1 (b)). Specifically, the data reveal that the concentration of CF₃SF₅  has grown from near zero in the late 1960s to ca 0.12 pptv in 1999 (or ca  2.5 × 10⁶ molecules cm^-3) with a current growth rate of ca  6 per cent per annum. Stratospheric profiles suggest that the lifetime of CF₃SF₅  in the atmosphere is between several hundred and a few thousand years. So is CF₃SF₅  a potentially serious greenhouse gas? 

The greenhouse effect

The greenhouse effect describes the trapping of infrared radiation, emitted by the Earth, which is absorbed in the atmosphere by small polyatomic molecules such as CO₂, CH₄ and H₂O. As a result the average temperature of the Earth is raised from 256 K, or -17°C, to a hospitable ~290 K. Often called the 'natural' greenhouse effect, this kept the Earth's temperature approximately static for hundreds of years up to the start of the Industrial Revolution, ca 1750.  

What we are concerned with today is an 'enhanced' greenhouse effect, caused by increases in concentration of greenhouse gases over the past 250 years. Nobody really doubts the scientific evidence that the concentration of the principal greenhouse gas, CO₂, has increased by ca 35 per cent over this period, from ~280 to ~380 parts per million by volume (ppmv), while the average temperature has also increased by ~1°C. What has not yet been proven is that there is a cause-and-effect correlation between these two facts. The consensus of world scientists, and certainly physical scientists,³ is that the CO₂ concentration and the temperature of the planet are strongly correlated, but there remains a small vociferous minority who believe otherwise. 

Although most attention has been given to CO₂ (and possibly CH₄ and H₂O), physical scientists now understand that there are larger polyatomic gases of low concentrations in the atmosphere which can contribute significantly to global warming. These gases, which include SF₆ and CF₃SF₅ ,^1,2 absorb infrared radiation strongly in regions where CO₂, CH₄ and H₂O do not absorb.  

Properties of greenhouse gases

In qualitative terms, there are two properties that are necessary for a molecule to be an effective greenhouse gas.  

• The molecule must absorb infrared radiation strongly in the black-body range of the Earth's emission, ca 5-25 μm, where CO₂, CH₄ etc  do not absorb. In practice, many C-F and C-Cl stretching vibrations around 10 μm contribute strongly. Such transitions are only observed if the vibration causes a change in dipole moment of the molecule. Figure 2  shows the calculated infrared absorption spectrum of CF₃ SF₅. The molecule has 24 vibrational modes, but only six have any significant infrared intensity. The four most intense bands occur in the atmospheric window 8-14 μm or 730-1260 cm^-1, where the major greenhouse gases CO₂, CH₄ and H₂O do not absorb. (Note that the vibrations of N₂ and O₂, comprising 99 per cent of the Earth's atmosphere, are infrared inactive.)  
• The molecule must have a long lifetime (at least 10 years) in the atmosphere; it must not be destroyed by photodissociation in the range 200-600 nm, and it must not react with the free radicals prevalent in the atmosphere.  

Furthermore, a greenhouse gas whose concentration is increasing rapidly owing to human activity will cause special concern. The gas CF₃SF₅ satisfies these criteria.  

Table 1 shows data for four greenhouse gases - CO₂ and CH₄ together cause ~70 per cent of the total radiation trapping; a chlorofluorocarbon CF₂Cl₂; and CF₃SF₅. The 'radiative forcing' is a measure of the strength of the infrared absorption bands over the range 5-25 μm, it is a per molecule microscopic property with units of W m^-2 per unit concentration. When multiplied by the change in concentration of pollutant over a defined time, usually 250 years from the start of the Industrial Revolution to the present day, the macroscopic radiative forcing in units of W m^-2 is obtained. We can then compare the effect of different pollutant molecules over a specific period. The greenhouse potential (GHP), sometimes called the global warming potential, measures the radiative forcing, A_x of pulse emission of a greenhouse gas, x, over a defined time, t, usually 100 years, relative to the time-integrated radiative forcing of a pulse emission of an equal mass of CO₂ (see equation (i)). 

The GHP is a dimensionless number that tells us how important one unit of mass (eg  1 kg) of a pollutant x is to the greenhouse effect compared with the same unit of mass of CO₂. The GHP of CO₂ is defined to be unity. Equation (i) can be simplified to give an analytical expression for the GHP of x over a time, t (see equation (ii)).⁴

The GHP of x reflects values for the microscopic radiative forcing, a_o, of greenhouse gases x and CO₂; the molecular weights of x and CO₂; the lifetime of x in the atmosphere τ_x; and the constant K CO₂ which can be determined for any value of t. K CO₂ has units of time, and is related (but not equal) to the lifetime of CO₂ in the atmosphere, the values of which range from 50 to 200 years.⁴

The macroscopic overall contribution of a pollutant to the greenhouse effect involves a complicated convolution of its concentration, lifetime and GHP value. Thus CO₂ and CH₄ contribute most to the greenhouse effect because of their high atmospheric concentration (note that the microscopic radiative forcing and GHP of both gases are relatively low). By contrast, CF₃ SF₅ has the highest microscopic radiative forcing of any known greenhouse gas (earning it the title 'super' greenhouse gas). (Absolute infrared absorption measurements have shown the microscopic radiative forcing of CF₃SF₅ to be 0.60 ± 0.03 W m^-2 ppbv^-1, which is even higher than that of SF₆.) The GHP of these two molecules is therefore very high - SF₆ is slightly higher because its atmospheric lifetime, 3200 years,⁴ is about three times greater than that of CF₃SF₅. However, the contribution of these two molecules to the overall greenhouse effect is still relatively small because their atmospheric concentrations, despite rising rapidly, are still low, at the level of parts per trillion by volume.  

Lifetime studies

The reactions that remove CF₃SF₅ from the atmosphere are important because they contribute to the molecule's lifetime and GHP value. The total removal rate per unit volume per unit time is given by equation (iii), where: 

• each of the five terms in the large bracket of equation (iii) is a pseudo-first-order rate constant; 
• [x] represents the concentration of species x, in this case •OH, O•, cations and electrons; 
• the first four terms represent reactions of CF₃SF₅ with •OH, O•, cations and electrons, respectively;  
• k_i  are the corresponding second-order bimolecular rate coefficients. 

OH radicals and electronically-excited O atoms, O•, are the most important oxidising free radicals in the atmosphere. The first term in the large bracket dominates in the troposphere (0 <altitude (h)<10km) ; the second term in the stratosphere (10 <h  <50 km); and the third and fourth terms in the mesosphere (h >50 km). In the fifth term, σλ and Jλ are the absorption cross section for CF₃ SF₅ and the solar flux, respectively, and Φ_λ is the quantum yield for dissociation at wavelength λ. In the troposphere, the summation for λ is over the range ca 290-700 nm, in the stratosphere ca  200-290 nm, and in the mesosphere the solar flux at 121.6 nm dominates all other vacuum ultraviolet (vuv) wavelengths. Equation (iii) assumes that the ion-molecule and electron attachment reactions lead to the removal of CF₃SF₅ by formation of dissociation products. Furthermore, secondary reactions of such products must not recycle CF₃SF₅.  

Experimental results

We measured, albeit in an indirect manner, the strength of the CF₃-SF₅ bond that connects the two radicals.⁵  We used tunable vuv radiation from the UK Daresbury synchrotron source to photoionise CF₃SF₅. We measured the translational kinetic energy released as CF₃SF₅⁺ dissociated into CF₃⁺ + SF₅ as a function of photon energy, and obtained the first dissociative ionisation energy of this molecule, ie Δ_r  H⁰ at 0 K for the reaction CF₃SF₅ → CF₃⁺ + SF₅ + e^- (see Fig 3). The error in each value of the kinetic energy is ca  20 per cent. A linear extrapolation to zero kinetic energy gave the first dissociative ionisation energy of CF₃SF₅ to be 12.9 ± 0.4 eV. We then determined the strength of the CF₃-SF₅ bond to be 3.86 ± 0.45 eV or 372 x 43 kJ mol^-1, and the enthalpy of formation of CF₃SF₅ at 0 K to be -1750 ± 47 kJ mol^-1. The high dissociation energy indicated a strong σ-bond, a slightly surprising result. This result confirmed, however, that uv photolysis in the stratosphere was unlikely to contribute to the rate of removal of CF₃SF₅ from the atmosphere.  

We have since made laboratory-based measurements relevant to the mesosphere, where ionic processes involving cations, anions, electrons and vuv photoexcitation at 121.6 nm dominate. We measured rate coefficients and product yields of small cations reacting with CF₃ SF₅ in a flow tube, the rate coefficient of low-energy electrons reacting with CF₃ SF₅, and the absorption cross section of CF₃SF₅ at 121.6 nm.⁴ Our results suggest that the dominant process that removes CF₃SF₅ from the mesosphere is low-energy electron attachment (ca  99 per cent), with vuv photodissociation only making a minor contribution (ca 1 per cent), see Fig 4. (The rate coefficient for electron attachment is measured as a function of mean electron energy, ε, in atmospheric pressure of N₂ (ε<0.5 eV) and Ar (ε>0.5 eV). The thermal rate coefficient at 298 K, where ε= 0.038 eV, is 7.7 ± 0.6 × 10^-8  cm³ molecule^-1 s^-1, and the sole product is dissociative, SF₅.^-6  Ion-molecule reactions make negligible contribution, not because the rate coefficients are too low but because the concentrations of the relevant ions in the mesosphere (eg N⁺, N₂⁺) in equation (iii) are too small. 

Physical chemistry data

To a physical chemist, the lifetime of a greenhouse gas is determined by the inverse of the pseudo first-order rate constant of the dominant chemical or photolytic process that removes the pollutant from the atmosphere.  

Consider, for example, CH₄. This gas is removed in the troposphere  via oxidation by the OH free radical: 

•OH + CH₄ H₂O + CH₃

The rate coefficient for this reaction at 298 K is 6.4 x 10^-15  cm³  molecule^-1  s^-1, so the lifetime is approximately equal to (k₂₉₈ [•OH])^-1. Assuming the tropospheric •OH concentration to be 0.1 pptv or 10⁶  molecules cm^-3, the lifetime of CH₄ is calculated to be ca  five years. This is within a factor of 2.4 of the accepted value of 12 years (Table 1). The difference arises because CH₄ is not emitted uniformly from the Earth's surface, a finite time is needed to transport CH₄ via convection and diffusion into the troposphere, and oxidation occurs at different altitudes in the troposphere where the  •OH concentration varies from its average value of 10⁶  molecules cm^-3.  

Table 1 Examples of four greenhouse gases and their contribution to global warming

We can regard this as an example of a two-step kinetic process, A → B → C, with first-order rate constants k₁  and k₂. The first step, A → B, represents the transport of the pollutant into the atmosphere, while the second step, B → C, represents the chemical or photolytic process (eg  reaction with an •OH radical in the troposphere, electron attachment in the mesosphere etc) that removes the pollutant from the atmosphere. In general, the overall rate of the process (whose inverse is the lifetime) will be a function of both k₁ and k₂, but its value will be dominated by the slower of the two steps. Thus, in writing the lifetime of CH₄ as (k₂₉₈ [•OH])^-1, we assume that the first step, transport into the region of the atmosphere where chemical reactions occurs, is very fast with k₁ k_2.

CF₃SF₅ behaves in the opposite sense, and now k_1≪k₂. The slow, rate-determining process is the first step, ie transport of the greenhouse gas from the surface of the Earth into the mesosphere, and the chemical or photolytic processes that remove CF₃SF₅ in the mesosphere will have little influence on the lifetime. We can define a chemical lifetime, τ_chemical, as (k_e [e^-] +  σ_121.6 J_121.6 Φ_121.6)^-1, the value of which will vary with altitude. In the troposphere, τ_chemical will be infinite because both the concentration of electrons and J_121.6 are effectively zero, but in the mesosphere τchemical will be much less. Assuming the electron attachment step is dominant, multiplication of k_e for CF₃SF₅ by a typical electron density in the mesosphere yields a chemical lifetime which is far too small and bears no relation to the true atmospheric lifetime, because most of the CF₃SF₅ does not reside in the mesosphere. Using calibration data for SF₆, the globally-averaged lifetime of ~1000 years for CF₃SF₅ (Table 2) comes from a weighted integration of the removal rates in the different regions of the atmosphere. Its lifetime is therefore determined by the meteorology that transports it into the mesosphere, and the chemical fate of each molecule when it reacts in that region with low-energy electrons; and radiation at 121.6 nm makes negligible contribution to the atmospheric lifetime. 

Table 2 Thermal electron attachment rate coefficients, absorption cross-sections at 121.6nm, and atmospheric lifetimes for CF₄, SF₆ and CF₃SF₅. From these data CF₃SF₅ appears to behave as a perturbed SF₆, and not as a perturbed CF₄ molecule

Generic lessons

Currently, there are no known undesirable chemical effects of low concentrations of CF₃SF₅ (and SF₆) in the atmosphere. However, their rapidly increasing concentrations and their exceptionally long lifetimes means that life on Earth may not be able to adapt to any changes these gases may cause in the future.  

It is worth remembering that the concentration of chlorofluorocarbons in the stratosphere grew from industrially-produced benign molecules to serious ozone-depleting molecules over a period of less than 20 years. Atmospheric scientists are thus investigating the photochemical properties of CF₃SF₅, which two years ago contributed just 0.003 per cent to the total radiation trapping, reasoning that small problems in this area have a tendency to become big problems.⁴ According to Professor Ravishankara,⁷ a physical chemist at the University of Colorado at Boulder, all such exceptionally long-lived molecules should be considered guilty unless proven otherwise.  

One solution would be to discover inert, dielectric gases with low GHP values which could be used as substitutes for SF₆ in industrial applications. Ring-based perfluorocarbons, such as cyclic-C₄F₈ and cyclic-C₅F₈ are possibilities. However, the simplest solution is that we should not put up into the atmosphere any more pollutants than are absolutely necessary. 

Richard Tuckett is professor of chemical physics in the school of chemistry at the University of Birmingham, Edgbaston, Birmingham B15 2TT.