The Atmospheric Chemistry of Hydrogen Cyanide (HCN)

Since 1981, three groups have reported spectroscopic detections and measurements of hydrogen cy- anide in the atmosphere. HCN concentrations (volume mixing ratios) of (1.5-1.7) x 10 -(cid:127)ø appear to characterize the stratosphere and the northern hemisphere's nonurban troposphere. In this paper we explore the atmospheric behavior of HCN by examining its chemical and photochemical properties. Its principal sinks are reactions with atmospheric OH and O((cid:127)D); precipitation appears to be a negligible sink. In the stratosphere, vacuum UV photons also attack HCN. Atmospheric model calculations show that HCN should be relatively well mixed in the troposphere and that its concentration decreases slowly with altitude in the stratosphere. Its atmospheric residence time appears to be about 2.5 years, although 1-5 years is a possible range. To maintain the observed atmospheric burden of HCN, an annual source of about 2 x 10 TM g nitrogen as HCN is required; we speculate as to the identity of these sources. Oxidation of HCN by OH, while the major sink for atmospheric HCN, is not simple or direct. Instead, oxidation proceeds from the HCN-OH adduct formed in HCN + OH reactions. These pathways and their uncertainties are outlined here.

and apply for 0øK except for HO 2 for which we took Howard's [1980] value of 2.5 _ 1 kcal/mol and for NCO we adopted 48 kcal/mol from Sullivan et al. [1983]. Note that all these potential reactions are endothermic (AHs > 0); rates of these reactions in the atmosphere would be negligibly slow. Data, 1928]. At the same temperature the Ostwald solubility coefficient (see, e.g., Wilhelm et al., 1977] is about 252 [Linke, 1958]. Further confirming data are found in the works by Lewis and Randall [1923] and Landolt-B6rnstein [1962]. On the basis of this solubility and the fact that HCN is a very weak acid HCNaq •-H + + CN-(K = 1.3 x 10 -9) one deduces that HCN has a very long atmospheric residence time against rainout. If 1 m of rain falls per unit area each year and it were saturated with HCN (with respect to, say, 200 ppt of gaseous HCN) it would take 34 years of average rainfall to remove the 200 ppt of HCN. This calculation is not very sensitive to rainfall pH; we took pH = 4 here. Let us proceed now to examine possible gas phase reactions that can affect the atmospheric behavior of HCN. We will conclude that the principal sinks for HCN are reactions with OH and O(•D) and in the upper stratosphere, photolysis. First we will examine and dismiss other potential reactions. Table 1 lists enthalpies of reaction, AHR, for potential reactions between HCN and 14 atmospheric gases. The significantly positive values of AHR shown in Table 1 indicate that each of these reactions would proceed very slowly if at all at atmospheric temperatures. Once again, the impression arises that HCN can be a relatively stable gas in the atmosphere, as was suggested by Crutzen et al. ['1979].

Contrary to extrapolations from information in several popular handbooks, HCN is not very soluble in water when low partial pressures (for example, < 1 torr) of HCN are in question. Henry's constant Kn = P/X, where P is the partial pressure (torr) of HCN and X is the mole fraction of HCN in the liquid, is Kn = 4000 at 18øC and 0.01 atm HCN [International Critical Tables of Numerical
It is possible to propose exothermic gas phase atmospheric reactions of HCN. Table 2 lists seven such reactants that branch into 14 sets of reaction products. Close inspection of molecular structures leads one to the conclusion that most of these reactions are unlikely to proceed in the atmosphere. For example, HCN +C10-•NCO+ HC1 would be a fourcentered reaction; four-centered reactions involving closed-electron shell reactants exhibit relatively large activation energies [see, e.g., Kneba and 'Wolfrum, 1980]. In other cases from Table 2, for example, HCN + O3-•  HOCN + 02, while one can imagine reasonable pathways   like   HCN + 03--} HOCN + 02 primary attacks on a closed electron shell molecule (HCN) are required. For example, the reaction of CO (isoelectronic with HCN) and 03 is unmeasureably slow (k < 10 -23 cm3/ molecule s). In several of the possible reactions listed in Table  2, extensive rearrangements are also required. Finally, there is no available evidence that any of the possible reactions in Table 2 actually occur except for one of the pathways with OH.
There are two likely atmospheric reactants of significance in the atmospheric chemistry of HCN: O(•D) and OH. Indeed, the reaction of OH and HCN has been studied in the laboratory [Phillips, 1978[Phillips, , 1979Fritz et al., 1983;R. Zellner and B. Fritz, unpublished manuscript, 1983]. From these laboratory experiments, it appears that an adduct is formed; the thermochemistry of the possible reaction pathways is shown in Table  2. The central values of AH298 ø for HCN, OH, CN, and H20 lead to the conclusion that HCN + OH--}CN + H20 is slightly endothermic (A/-/a = + 1.5 kcal/mol). Further, for a hydrogen abstraction reaction like this, BEBO calculations indicate that there should be an activation energy of at least 9 kcal/mol (R. Zellner and B. Fritz, unpublished manuscript, 1983). The experiments of Fritz et al. [1983] imply that this reaction path is not followed at atmospheric temperatures. Further, the exchange reaction path (OH + HCN HOCN + H) which has been suggested to be one of the dominant HCN oxidation routes in flames [Haynes, 1977] Table 1 for definitions and sources of thermodynamic data; a heat of formation, AH s, of 9 kcal/mol has been taken for CHO [Benson, 1976]. We have taken AHs > -14 and > -8 kcal/mol for HNCO and HOCN, respectively [Sullivan et al., 1983;Benson, 1978]. Although exothermic, most of these reactions are concluded to be highly unlikely as discussed in the text. The only likely reactions affecting atmospheric HCN is with OH in which an adduct is formed and reaction with O(•D). We cannot dismiss this possibility, but we consider it unlikely that this bimolecular abstraction reaction can compete with termolecular O2 addition, especially in the troposphere. To compete successfully, ka must exceed the effective second-order rate constant, k., for the O2 addition shown at point A in  Table 3 lists six such reactions and their thermochemistry. Unfortunately, reaction rates are not available for all these reactions especially at atmospheric temperatures. The reactions of Table 3 are not fast; some are very slow. The reaction of C2H4OH most closely resembles our candidate reaction both in complexity and exothermicity (for HCNOH + O2-} HOCN + HO2, AHR = -12.5 kcal/mol). Table 3   One can also imagine that OH or CO could compete with NO at this point. At point C in Figure 1, the proposed attack by 03 is by analogy with NH 2 + O3--} products, a reaction that is found to be relatively fast [Hack et al, 1982] and whose products are suggested to be NH20 + 02. Also, at point F in Figure 1

ATMOSPHERIC MODEL CALCULATIONS AND DISCUSSION
By performing calculations with an atmospheric photochemical/transport model, we can learn several things about atmospheric HCN. For example, by assuming steady states for its atmospheric vertical profile and for its ground level sources, we can deduce its atmospheric residence time. Further, we can estimate the total size of its sources by summing the loss rates for HCN at all altitudes. Individual HCN sources (discussed below) should, of course, sum to the total source estimate.
The numerical model we used for these calculations is very similar to that described by Cicerone [1979]. For HCN, we adopted a fixed-density lower boundary condition at (z = 0 km) equivalent to a volume mixing ratio of 1.7 x 10-•ø. At

the model upper boundary (80 km) we employed the condition that the upward flux, •p(80) of HCN equals the steady state loss above that altitude. This flux was calculated from c) =fMH(J + k,[OH] + k2EO('D)]) (2) where the term in parentheses is the sum of the first-order loss processes for HCN (in s-•) due to photolysis (J), reaction with OH (k•[OH])and with O(•D)(k2[O•D]). H and f are the vertical scale height and the volume mixing ratio of HCN at 80 km, respectively. The three loss processes for H CN were
included in the model at all altitudes, z, from zero to 80 km.

OH, O(•D), J k•(z), km cm-3 cm-3 s-• cm 3/molecule s
Actually, Coffey et al. reported a total column amount at all altitudes above 12 km; we have indicated that their deduced mixing ratio applies to altitudes below 40 km because only about 1% of the measured absorption could arise from over 40 km even if HCN were distributed uniformly. In other words, their data contain virtually no information on the HCN above 40 km, so comparison with our calculated profile is meaningless for altitudes above, say, 40 km. Our model does predict a slow but significant decrease of HCN above 20 km. The Carli et al. [1982] data were reported as HCN mixing ratios of 1.3 to 2.6 x 10-•o at all altitudes between 20 and 40 km. With OH densities 50% those of Table 4, HCN decreases more slowly with altitude; at 40 km it is 1.13 times that shown in Figure 3.
Attack by OH (with subsequent rapid oxidation to CO + NO, as shown in Figures 1 and 2) Table 4, twice those of Table 4, and half those of Table 4 we find ,• = 2.5, 1.3, and 5.0 years, respectively. From this estimate of atmospheric residence time, we may proceed to predict southern hemispheric (SH) tropospheric mixing ratios of HeN which are as yet unmeasured. By assuming that the sources of HeN are primarily in the NH (see below) and applying the analysis that Khalil and Rasmussen [1983] used for CH4 (whose principal sink is also OH) we estimate that the ratio of northern hemispheric HeN mixing ratio to that in the SH is  What is the origin of atmospheric HeN? According to our calculations, an annual source of (1-3) x 10 TM g(N) as HeN is required to maintain the measured atmospheric HeN burden. This annual source of HeN to the atmosphere is less than 1% of the estimated global annual amount of N in NO,, [Logan, 1983] from all sources. Significant indirect sources such as the atmospheric decomposition of R-CN species (e.g., CH3CN ) are very unlikely as discussed above. Also, in situ injections by jet aircraft are apparently insignificant [Robertson et al., 1979] because with adequate O2, turbine engines produce very little HeN. Some catalytic converters that reduce automotive emissions of NO,,, when in ill repair, are known to produce high concentrations (500 ppm) of HeN. The possibility that HeN is produced by lightning has been considered by Chameides and Walker [1981] and J. T. Kasting (private communication, 1981). With present epoch oxygen levels, this HeN source appears negligible, although it should be noted that emissions from CN radicals are observed occasionally from lightning-disturbed air [Wallace, 1960;Salanave et al., 1962]. Similarly, thermodynamic equilibrium calculations indicate that volcanoes can emit HeN, especially in high-temperature, low-oxygen zones (R. Prinn, private communication, 1982), but this source is difficult to quantify, and it is probably relatively small. Direct evident does exist for the release of HeN from several biological sources and from several industrial practices such as the production of coke. The production and release of HeN from higher plants and from bacteria and fungi is well documented [Conn, 1980;Bach, 1948;Robbins et al., 1950;Marshall and Hutchinson, 1970], although no estimates of global release rates into the atmosphere are available. These biogenic HeN sources clearly need further investigation. Returning to nonbiogenic sources, large amounts of HeN appear in coke oven gas when coke is produced [Grosick and Kovacic, 1981] and in other coal carbonization processes. Generally, high-temperature, low-oxygen processes are likely to produce HeN; an example might be the process of steel degassing when Argon is used in place of oxygen as a purge gas [Cottrell, 1967]. Many other hydrocarbon-rich flames are throught to produce HeN, but little if any HeN escapes oxidation to NO in general [Glassman, 1977]. Smoldering biomass might produce HeN and CH3CN; this should be explored. Thus, in principle, several likely sources exist for atmospheric HeN. Biological sources ivould be of great interest to quantify to determine if atmospheric HCN is mostly natural or anthropogenic.