NOTES
722 spond to a weight change of 60 pg. However, just because the melt i s so dilute in oxide, very small and undetected changef in weight will correspond to very large changes in the partial pressure of COz. It follows from the above argument that the dynamic method is not suitable for the measurement of liquid salt decomposdion pressures if the salt and oxide are miscibIe-a condition which usually applies. For the decomposition of solid salts which do not form solid solutione with their oxides the dynamic method may have the advantage of speed over the static method. Moreover the good agreement between the observed dissociation pressures of calcium carbonate1 and those calculated from thermal data3 indicates that the assumption of thermodynamic equilibrium holds even if the fraction of condensed phase present as oxide is extremely small. The hazard in the dynamic method i s that one depends on kinetic factors to prevent the complete conversion of the salt into the oxide. I n the static method, on the other hand, the ratio of the two solid p h a w is eiontrolled by the experimenter who can prove that equilibrium has been attained by showing that the gas prt-mure is independent of that ratio. For this cam she mleaning of the term “decomposition pressure” i s unambiguous since the activities in eq 1are unity. If the salt a i d oxide do form a solid solutiona question to be invwtigated experimentally in each case--the equ ilibriurn pressure will depend on the solid phase cornposition, and there is no longer a unique value of the “decomposition pressure’’ at each temperature. Decompasibion pressures measured above one-phase liquid mixtures by the static method are also not unique. However, 1his method permits the experimenter to determine the composition of the melt in equilibrium with gas of Bnown pressure. From this information and a knowledge ( d activity coefficients (or the assumption of ideal behavior) K , and the thermodynamic variables derived from i t s tennperature dependence can be calculated. (For ,an example of such a study see ref 4.) However, the term “decomposition pressure” should not be used for these systems, since it conveys the impression that it is a unique function of temperature only. A meaningful description of the equilibrium state requires the statement of the equilibrium pressure above a melt of specified composition and temperature. I n sumlxlar.jl, the dynamic method cannot give meaningful resuits for single-phase liquid mixtures, and its use for measuring Ihe decomposition pressure above a two-phase solid system is hazardous if the relative amount of one of these phases is very small and cannot be controlled. Finally, the generality of the above arguments makes them applicable to any reversible process of the type condensed phAse 1 == condensed phase 2 gas(es).
+
(4) R. F. Bartholemew,J . Ph,ys. Chem., 70, 3442 (1966). The Journal of l % y s b l Chemiatry, Vol. 76, No. 6 , 1971
Mass Spectrometric Study of the Reaction of Hydrogen Atoms with Nitrosyl Chloride
by 1L1. R. D u m , M. N ! . Sutton, C. G. Freeman, &J. !IMcEwan, , and L. F. Phillips* Chemistry Department, University of Canterburg, Christchurch, New Zealand (Received October 26, 1970) Publication costs borne completely bg The Journal of Physical Chemistry
The reaction of H atoms with ONCl has been studied previously by Clyne and Stedman,’ who found that the products were HC1 and NO and that the stoichiometry of the reaction was close to 1 : 1in a flow system in which the walls were coated with phosphoric acid. They also showed that the primary reaction H
+ ONCl --+HC1 -I- NO
(1)
was very fast (IC1 > 1.5 X 1Q-12 cm3 molecule-’ sec-l) and suggested that the reaction could be used as a gasphase titration for the measurement of H atom concentrations. We have investigated the reaction mam spectrometrically in a fast-flow system at pressures near 0.1 Torr. We used sufficiently small reagent concentrations to allow the progress of the primary reaction to be followed over a range of several milliseconds and thus to allow measurement of the primary reaction rate. I n addition, for reaction times of 100 msec or more, we have confirmed the results of Clyne and Stedman regarding the nature of the reaction products and overall stoichiometry in a system in which the walls are poisoned against recombination of chlorine atoms. However, when the walls are not freshly poisoned, we find that the number of hydrogen atoms removed by each ONCl molecule typically rises to a value of about 2, and varies in a nonreproducible manner as the wall coating ages. This behavior we attribute to the presence of a chain reaction, involving chlorine atoms, which is initiated by the reaction of R with vibrationally excited HCI produced in the primary reaction. Provided chlorine atoms are removed only by reacting with ONC1, and not by recombination on the wall, the chain does not affect the overall stoichiometry. Hence, if the reaction is to be used to estimate hydrogen atom concentration on the basis of an assumed 1:l stoichiometry, care must be taken to ensure that an effective wall coating is present. Experimental Section The apparatus was similar to that used in our previous studies oi reactions of atoms with CL0.2 The flow (1) M. A. A. Clyne and D. H. Stedman, Trans. Faraday Soc., 62, 2164 (1966). (2) C. G . Freeman and L. F. Phillips, J . Phys. Chem., 72, 3025 (1968).
923
NOTES
speed in the 1 7 m m i d . reaction tube was in the range 8.5-10.6 m sec-l, except when the gas flow was throttled to allow measi~reme~its to be made at long reaction times. For must of the experiments, including all of the primary rate measurements, the walls of the flow tube were poisoned with phosphoric acid.3 Hydrogen atoms were generated by passing a stream of argon containing 1-10% of Nzthrough a microwave discharge; their concentration w , calculated ~ from the decrease in peak height a t mass 46 which resulted from the titration reaction with an excess of NO2.* Typically 80% of the E&was diasociated to atoms, so that vibrationally excited Hz should have been relatively unimportant in this system. M a theson nitrosyl chloride was purified by conclensatiori onto Pz06followed by repeated fractionation with a LeRoy still. Other materials were as used previously.2 All measurements were made at room temperature (ca. 23').
Results and ~~~~~~$~~~ The electron-impact, mass spect>rumof ONCl appears not to have been reported previously, so we present it in Table I. The main ieature of this spectrum is the very low contribution from the parent ONC1+ peak. The parent peak is a1:m very small or absent in the mass spec~ impact.6 Because of the lack trum obtained b . photon of a significant paren!.. peak we have been obliged to monitor the OKGI concentration using the NC1+ peak at mass 49; unfortunately the larger NO+ peak could not be used because NO is SL product of the reaction and the sensitivity of the instrument at mass 30 is almost the same for NO and ONCI. The low sensitivit,y of the mass spectrometer for ONCll at mass 49 wa8 a major source of difficulty during
Table I: Mass Spectrum of Nitrosyl Chloride",b Ma66 no.
Species
14
N"
16 30 31 35
0"
37
@I + NO2 +
46 49 51 53 65 67
70 72 74
NO + 'bMO +
ci
+
N85SC:I +
N W ~+,0 037C:I +
0~35c1+
ow 87131 " liis36(3,
+
3b I 877121 4-
+
37 ~ 3 7 ( > 1 ~
3 a +
Relative abundance
3.00 0.85 100.0 0.38 18.7 5.9 0.23 1.33 0.59 0.05 0.03 0.01 1.00 0.62 0.10
a Source pressure. 9 X 10-8 Torr; electron energy: 50 1'. The peak a t mass 46 can be attributed to the presence of ca. 0.57, NO%; those a t masses 70, 72, and 74, arise from a similar concentration of chlorine impurity.
the measurements of the primary rate, and in fact these measurements could not be attempted at all except when the instrument was behaving exceptionally well. The mean of 28 determinations of k,, calculated on the assumption of 1:1stoichiometry a t short reaction times, was 3.0 X 10-l' cm3 molecule-', with a standard devi% tion of 1.0 X lo-". Of these results, two sets of six determinations each were made under what appeared to be optimum conditions and were considered particularly reliable. The mean of these 12 values gave k1 = 2.7 X lo-" cm3 molecule-' sec-I, with a standard deviation of 0.5 X 10-l'. We give this as our preferred final value for kl, with an estimated error of 1.5 times the standard deviation. One of the two sets of six measurements was obtained with excess ONCP, the other with excess H atoms. The partial pressure of the reagent which was in excess was typically 3 X Torr. Recently, Niki, et aE.,' reported the results of a mass spectrometric study which gave kl = 4.5 molecule-' sec-l, in fair agreement with our value. (According to a referee of this paper the value quoted at the 160th National Meeting of the American Chemical Society was (3.2 f 1.6) X in which case the agreement is very satisfactory indeed.) The stoichiometry of the reaction was determined by measuring the number of ONC1 molecules removed by a known concentration of atomic hydrogen, at reaction times greater than 100 msec, with excess OWCl present. Measurements were made with both old and fresh coatings of phosphoric acid, and with uncoated Pyrex walls. I n the freshly-coated system the number of El atoms removed per ONCl molecule consumed was found to be 1.02, with a standard deviation of 0.08, in good agreement with CIyne and Stedman's figure of 0.94 (standard deviation 0.86) for the same quantity. However, when the walls were not freshly coated the number of H atoms removed per ONCl molecuIe destroyed ranged from 1.2 to 2.3, with ti tendency for higher values to be obtained as the age of the coating increased. This implies that the observed sto~ch~ometry is dependent upon secondary reactions imolving atomic chlorine, since the different results were obtained under conditions where different numbers of chlorine atoms could be lost by wall recombination. ( tion of hydrogen atoms is a slow process in both coated and uncoated Pyrex systems.) The HC1 molecule produced by reaction P is vibrationally excited, molecules in levels up to v = 9 having (3) E. A. Ogryzlo, Can. J. Chem., 39, 2656 (1961). (4) E. F. Phillips and H. I. Schiff, J. Chem. Phys.. 37, 1233 (1962). (5) R. F. Ileidner and J. V. V. Kasper, ibid., 51, 4163 (1969). (6) A. J. Nicholson, G.S.I.R.O. Melbourne, private communication. (7) H. Niki, D. H. Stedman, and D. Steffenson, Abstracts, 160th Kational Meeting of the American Chemical Society, Chicago, Ill.,
Sept 1970, p 116. The Journal of Physical Chemistry, Vd.76, No. 6,1971
NQTES
724 been detected from their infrared emission.8 Hence the secondary reaction
-t HCI--+ Ha
+ C1
(21 would be expected ito occur up to 100times faster for the C1 molecules than in the thermallyThe factor of 100 is derived on the assumption that all of the necessary 3 kcal mol-' of activation energy1 can be accepted in the form of vibrational exci-tation of the WCl. With this assumption the effective value of kl!is about 1.5 X 10-l2 cmSmolecule-' sec-I. I n tli~efreshly coated system the chlorine atom produced by reaction 2 would react with ONCl according: to
C1 + ONC1 ----P Clz + NO Thus preserving the 1:1 stoichiometry. reaction .f Cl2 --+ HC1
+ C1
(3) The fast
C1
+ wall --+l/zClz
(5)
which, together with reaction 4, would bring about a rapid catalyzed removal of H atoms. I n the limit of ks >> ks[ONCI] the result would be to remove three H atoms per ONCl molecule destroyed. Thus this mechanism can account satisfactorily for the observed stoichiometry in both the poisoned and unpoisoned systems. I n view of the small amount of undissociated Hz emerging from our microwave discharge, the reaction C1
(4)
cma molecule-l sec-' lJ) would then be expected to compete effectively with reaction 1, except in thch presence of a large excess of ONC1. This view is supported by the observation that less than O . l . ~ $ of the ONCI appeared in the form of Clz in the products. Reactions 3 and 4 constitute a chain which would continue until all of either the ONCl or the atomic hydrogen had been used up, and which would give rise to the observed 1: 1 stoichiometry. The presence of this chain does not affect our measurement of the primary rate because our value of is ten times as
The Journal Ojp PhgsCal Chemistry, Vol. 76, NO.6,1971
large m the effective value of kz, so that reaction 2 would not be significant at short reaction times with very low reagent pressures. I n a clean Pyrex system, or with an ineffective wall poison, reaction 3 would experience competition from the process
+ HZ(V>> Oj --+
HC1
+H
(6)
+
which Niki, et al., found to be important in the H Clz system is not IikeIy to have been significant except perhaps when H atoms were in very large excess. This reaction could not affect our primary rate measurements because it requires the prior occurrence of reaction 2. Acknowledgments. This work was supported by the New Zealand Universities Research Committee and by Grant AF-AFOSR-1265-67 from the United States Air Force Office of Scientific Research. (8) J. K.Cashion and J. C . Polanyi, J . Chem. Phys., 35, 600
(1961).
