Office of Special Assistant to Deputy Secretary of Defense
for Gulf War Illnesses Update

Prepared by

Richard McNally and John E. Cockayne
Science Applications International Corp.
2111 Eisenhower Avenue, Suite 200
Alexandria, VA 22314
Tel (703)683-7900


This technical note describes the estimation of atmospheric decay of agents sarin and cyclosarin. There is always some concentration of radicals in the air that can initiate changes in GB and GF chemical warfare agents, which rapidly cascade into a reduction in toxicity. The hydroxyl radical (OH) is the radical of concern for the daylight hours following the Khamisiyah pit incident; nitrate (NO3) is the radical of interest during nighttime hours. The decrease of toxicity is many factors of ten for a hydroxylation, because the altered GB (sarin) and GF (cyclosarin) molecules and byproducts can no longer bind appropriately to affect the reactions within nerve junctions; that is, at the neuron/neuron and neuron/muscle (neuromuscular) junctions or synapses.

The hydroxyl radical produced in moist air when short UV light is present is very reactive, with a e-folding time scale limited to seconds. A probabilistic estimate of hydroxylation on G-agents shows that the reaction produces a minimal likelihood of depleting the OH concentration. However, there is a minimum to maximum range factor of over two in estimated decay rates that can be applied to the Khamisiyah case, because the concentration of the hydroxyl radical or any correlative chemical was not measured along the actual trajectories of the dispersing agent vapors. The Khamisiyah Pit incident is discussed in this paper, along with recommended reasonable upper bounds of half-lives and time scales (i.e., leading to higher, more conservation concentration estimates) and associated lower bounds of reactivity rates.


The GB, GD, and GF nerve agents are derivatives of phosphoric acid with its three hydroxyls replaced with a fluorine atom, a methyl group, and a hydrocarbon ester replacement for the third hydroxyl, which does not directly modify the core P-O bond. This differentiating ester group is isopropyl (C3H7) for GB, pinacolyl (C6H13) for GD and cyclohexyl (C6H11) for GF, all of which have so-called saturated hydrocarbons. The net result is that the two different bonds for (1) the ester's P-O link and (2) any H-atom to a secondary carbon atom are vulnerable to breakage by the OH or other aggressive ambient radicals, such as nitrate.

The structural similarity of the phosphate's ester group is inferred when using triethyl and trimethyl phosphates (TEP and TMP) as simulants for chemical changes. Bunnett (1995) selected TEP (i.e., (OC2H5)3P) as a nominal surrogate for the hydroxyl reaction rate constants of the nerve agents GA, GB, GD and VX; they all have the P-atom bonded to an oxygen or sulfur atom that is also bonded to a multi-carbon group, which TMP does not have per se. The TEP G-surrogate for OH-reactivity is one of the few organophosphorus compounds with a published experimental reactivity rate (Atkinson, 1989), which is 5.5 x 10-11 cm3/molecule-s. For a GWI dosage perspective of this bimolecular rate, an average annual sunny daytime OH· concentration of 2 x 10+6 radicals/cm3 (Prinn et al., 1995 and Tanner et al., 1997) results in an average daytime e-folding time scale of 2.5 hours. That corresponds to a half-life of only 1.75 hours (i.e., 105 minutes); the precision should be ignored because of the single-digit precision for OH. When averaged over the full 24 hours, the half-life becomes an effective 3.5 hours for the TEP G-surrogate for OH reactivity.

Converse to Bunnett's choice of TEP, the TMP's ester variation was used as an OH -reaction simulant in an experimental program to obtain the GD reaction rate (i.e., C6H13 ester (Segers et al., 1990)). In that effort, the stated emphasis was on the phosphorus moiety, without apparent motivation except for the contract sponsor's additional interest in the VX reaction rate and the rate's simulation in establishing that the OH -reactivity testing technique was working correctly. (The VX reactivity rate was not obtained in the experimental system due to its low volatility.) Because the Segers et al. (1990) work started in early 1990, the team should have had an opportunity to find that the TEP's reactivity rate was also measured previously and available presumably to verify a properly working test chamber (Atkinson et al., 1988). Note that this is the source for the above cited Atkinson (1989) reference for TEP reactivity, which also included the TMP value.

The apparent simplicity of the covalent bonds also permits the use of an empirical modeling technique to estimate the gas-phase hydroxyl reaction rate constant at nominal temperatures. The Atmospheric Oxidation Rate Program (AOP) operated by Syracuse Research Corporation (Meylan and Howard, 1993) produces an OH reactivity rate value of 4.02 x 10-11 cm3/molecule-s (personal communication from Bill Meylan to John Cockayne, 25 May 1999). The AOP computer code uses the empirically-based methods developed since the mid-1980s by Dr. R. Atkinson and co-workers at the University of California at Riverside, which are better known as the Structure-Reactivity Relationships (Kwok and Atkinson, 1995).

The AOP calculated 4 x 10-11 cm3/molecules-s reactivity rate was independently estimated by Dr. Atkinson, which presumably verifies the AOP coding, and was communicated previously to others involved in the G-agent hazard assessments (personal communication from Roger Atkinson to John Cockayne, 22 May 1999). Nevertheless, a range of accuracy is unknown for this value from an empirical method. This bimolecular rate produces an average daytime 2.4-hour half-life for direct comparison to the above 1.75-hour half-life for one simplified surrogate. From the empirical methodology, the fluorinated ethers data set with the strong fluorine bond effect suggests significantly smaller rates than calculated formally, which makes the 2.4 hours a minimum half-life. Assuming a factor of two effect from the GB's fluorine atom, the net result would be a 9.6 hour half-life average over a full 24-hour period.

In support of using a factor of two slower reactivity, note that the subtle sensitivity for fluorine bonding is implied partially by the ratio of measured rates for TEP and TMP, using the same general experimental setup used by others in the reactivity research group under R. Atkinson. The TMP reactivity rate for hydroxylation is 7.5 times smaller than the above stated TEP reactivity rate; the latter "large" reactivity rate resulting apparently from adding/inserting the CH2 within each of the three OCH3 groups. Thus, changing from the ethyl moiety of TEP to the isopropyl moiety of GB should be anticipated to increase the vulnerability for H-atom abstraction during hydroxylation to a reactivity rate above the 4 x 10-11 cm3/molecule-s from the AOP. In summary, with regard to reactivity rate variations and uncertainties in the empirical Structure-Reactivity Relationships modeling, Kwok and Atkinson (1995) state:

"The present estimation technique is reasonably reliable when used within the database used in its derivation …Thus, while the present estimation method appears to be reliable to well within a factor of 2 for…reactions with alkanes and alkenes…, the calculation…for organic compound classes other than the alkanes and alkenes is prone to significant error."

The relative accuracy of the AOP is assumed to be adequate for scaling the reactivity rate constant from GB to GF (i.e., from the C3H7 to C6H11 ester). That is, the relatively more H-abstraction vulnerable structure of the cyclohexyl leads to the e-folding time scale of GF being only 60% of that of GB for the same OH (personal communication from Bill Meylan to John Cockayne, May 1999). The degradation rate impact of this factor is in a power or exponent. However, the net result is complicated by the criterion for human response, adjusted to the pertinent low levels, that is a factor of three lower for GF compared to GB (Mioduszewski et al., 1998).

The normal hydroxyl concentration is approximately a sine function variation over p radians, just between sunrise and sunset, from short UV (wavelength < 320nm) sunlight effects on ozone and a subsequent single oxygen atom merging into H2O and a splitting into two OH radicals. The D-state oxygen atom and H2O reactivity rate of 22 x 10-11 cm3/molecule-s competes with air deactivation of the single oxygen atom to its P-state. In the first field program with independent measurements of the instantaneous OH·, Tanner et al. (1997) show a time series on a Fall equinox date where the peak OH· is ~ x 10+6 radicals/cm3 in Colorado. Such behavior results in an average over 24 hours of 1 x 10+6 radicals/cm3, which is the same value deduced from the long time scale of the trichloroethane pollutant involved in the depletion of ozone (Prinn et al., 1995).

The nitrate situation for nighttime degradation depends directly on pollution to build up NO3 to where its effect is significant. Since there are no nitrate measurements in the region of interest, nighttime agent degradation due to nitration was simply ignored as a conservative assumption.


The diurnal variation of the hydroxyl radicals concentration can be represented as x sin(t/12) x 10+6 radicals/cm3, where t represents the hours after an 0600 Gulf time sunrise; and zero for 1800 to 0600 Gulf time (i.e., during nighttime). The maximum hydroxyl reactivity rate constant for GB is £ 4 x 10-11 cm3/molecule-s. Combined, these two factors give a time-varying half-life in hours of 1.5 / sin(t/12), or an e-folding time constant of 2.2 / sin(t/12) hours, which goes to infinity for the other 12 hours (nighttime) of each day. The reasonable upper bounding case is a two times longer time constant and half-life because of the qualitative empirical trends from the above TEP baseline to (1) the GD's reactivity rate constant (note 6 "ester" C-atoms), (2) the TMP's measured much lower reactivity rate constant, and (3) an optically thin layer of aerosols after recent rain showers at and around Khamisiyah.

The specific variation by hour following the event is plotted and summarized as follows:

The latter higher value was obtained using the AOP ratio of reactivities. For reference, the actual AOP (v1.89) estimated values, including two other normal nerve agents, are:

  Agent Hydrogen abstraction rate
  GB 40.2406 x 10-12 cm3/molecule-s
  GD 49.7222 x 10-12 cm3/molecule-s
  GF 67.3693 x 10-12 cm3/molecule-s
  VX 90.4382 x10-12 cm3/molecule-s

Figure A-22. Pie chart of Dugway tests showing partitioning of the source term


Atkinson, R., S.M. Aschmann, M.A. Goodman, and A.M. Winer. 1988. Kinetics of the gas-phase reactions of the OH radical with (C2H5O)3PO and (CH3O)2P(S)Cl at 296� 2 K. Int. J. Chem. Kinet., 20, 273-281.

Atkinson, R. 1989. Kinetics and Mechanisms of the Gas-Phase Reactions of the Hydroxyl Radical with Organic Compounds. Monograph No. 1, J. of Phys and Chem Ref Data.

Bunnett, J.F. 1995. Some Problems in the Destruction of Chemical Munitions, and Recommendations Toward their Amelioration (Technical Report). Pure and Appl Chem, 67 (5), 841-858.

Kwok, E.S.C., and R. Atkinson. 1995. Estimation of hydroxyl radical rate constants for gas-phase organic compounds using a structure-reactivity relationship: an update. Atmos. Environ., 29, 1685-95.

Meylan, W.M., and P.H. Howard. 1993. Computer estimation of the atmospheric gas-phase reaction rate of organic compounds with hydroxyl radicals and ozone. Chemosphere, 26, 12, pp. 2293-2300.

Prinn, R.G., R.F. Weiss, B.R. Miller, J. Huang, F.N. Alyea, D.M. Cunnold, P.J. Fraser, D.E. Hartley, and P.G. Simmonds. 1995. Atmospheric trends and lifetime of CH3CCl3 and global OH concentrations. Science, 269, pp. 187-192 (14 July 1995).

Segers, D.P., R.B. Spafford, and R.B. Watkins. 1990. Specialized Laboratory Support: Atmospheric Chemistry of CW Agents, Volume I: Main Report. SRI-APC-90-1038-6819-002.

Tanner, D.J., A. Jefferson, and F.L. Eisele. 1997. Selected ion chemical ionization mass spectrometric measurement of OH. J. of Geophys Res, 102, D5, pp. 6415-6425.

Tuazon, E.C., R. Atkinson, S.M. Aschmann, J. Arey, A.M. Winer and J.N. Pitts, Jr. 1986. Atmospheric loss processes of 1,2-dibromo-3-chloropropane and trimethyl phosphate. Envirion. Sci. Technol., 20(10), 1043-1046.

| First Page | Prev Page | Next Page |