RATES OF THE REACTIONS CN+H2CO AND NCO+H2CO IN THE TEMPERATURE-RANGE 294-769 K

(1)

Chemical

Physics

E L S E V I E R Chemical Physics 200 (1995) 431-437

Rates of the reactions CN

+ H 2 C O

and NCO + H2CO in the

temperature range 294-769 K

Yu-Wei Chang, Niann S. Wang *

Department of Applied Chemistry, Chiao Tung University, Hsinchu, 30050, Taiwan, ROC Received 14 March 1995; revised 18 July 1995

Abstract

The rate coefficients of the reactions: (1) CN + H2CO ~ products, and (2) NCO + H 2 C O ---* products in the temperature range 294-769 K have been determined by means of the laser photolysis-laser induced fluorescence technique. Our measurements show that reaction (1) is rapid: k1(294 K ) = (1.64 + 0.25)X 10 - t l cm 3 molecule - t s - l ; the Arrhenius relation was determined as kl = (6.7 + 1.0) × 10-11 exp[( - 412 + 20)/T ] c m 3 molecule- 1 s- 1. Reaction (2) is approximately a tenth as rapid as reaction (1) and the temperature dependence of k 2 does not conform to the Arrhenius form: k 2 = 4.62 × 1 0 - 1 7 T 1'71 exp(198/T) c m 3 molecule-1 s-1. Our values are in reasonable agreement with the only reported measurement of kl; the rate coefficients for reaction (2) have not been previously reported.

1. Introduction

H C N is important in the ignition of nitramines and combustion of fuel-bound nitrogen [1-3]. CN and N C O radicals are two intermediates produced during oxidation of HCN. The N C O radical is also a key intermediate in a process R A P R E N O x (rapid reduction of nitric oxides), [4,5] to treat exhaust gases to decrease nitrogen oxides emitted from internal combustion engines. H 2 C O is a product of ther- mal decomposition of composite propellant R D X (1,3,5-trinitro-l,3,5-triazacyclohexane) [1]. We have investigated the kinetics of CN and N C O radicals with H 2CO to assess their importance in the combus-

* Corresponding author.

tion system of nitrogen compounds, according to the following reactions:

CN + H2CO ~ H C N + HCO, (1)

NCO + H2CO ~ H N C O + HCO. (2) Yu et al. [6] investigated the kinetics of reaction (1) using the laser photolysis-laser induced fluorescence technique. They observed non-Arrhenius be- havior for reaction (1) between 297 and 673 K: k t = 2.82 × T 2'72 e x p ( 7 1 8 / T ) cm 3 molecule- 1 s - 1 with k 1 = ( 1 . 6 6 _ + 0 . 0 3 ) × 1 0 -11 c m 3 molecule - t s -1 at 297 K. Calculations for a loose transition structure yielded reasonable agreement with their experimental values. Tsang [7] estimated k 1 = 7 x 10 -11 c m 3 molecule -1 s -1 based on the rate coefficient of abstraction of an H atom by CN from propane [8].

No kinetic measurement of reaction (2) has been previously reported. Only the value k 2 -- 1 × 10 -11

(2)

432 Y.-W. Chang, N.S. Wang/Chemical Physics 200 (1995) 431-437

cm 3 molecule -1 s -~ with a negligible activation energy was estimated by Tsang [7].

In continuing the kinetic measurements on reactions of CN and NCO, needed to address the rele- vance of free radical reactions in the combustion chemistry of nitrogen compounds, we employed the laser photolysis (LP)-laser induced fluorescence (LIF) technique to investigate the kinetics of the title reactions in the temperature range of 294-769 K and with He at a pressure of 52-201 Torr.

2. Experimental section

The LP-LIF system employed in this work has been previously described [9]. The resistively heated reaction cell is a six-armed quartz flask with a volume of about 250 ml. Two pairs of baffled side arms fitted with Brewster windows are mounted mutually perpendicular and they are used to admit the dissociation and probing laser beams that cross at the center of the reactor. The arms are flushed with a small flow of He to avoid accumulation of photolytic products on the windows. The fluorescent signal was collected with a photomultiplier tube (Hamamatsu R955 or R212UH) through a set of lenses and an interference filter from the side window perpendicular to the laser beams. The LIF signal was amplified (Hamamatsu C1053-01,5 MHz) and averaged with a gated integrator (Stanford Research System, SR250). A microcomputer was used to store and to analyze the collected data.

CN radicals were produced by photolysis of ICN at 248 nm (KrF) or BrCN molecules at 193 nm (ArF) with an excimer laser (Lambda Physik LEX- TRA 50). The concentration of CN radicals was monitored with a dye laser (Continuum ND60) at 387.63 nm pumped by a N d - Y A G laser (Continuum NY61) with a dye Exalite 389, which excited CN radicals from the X 2 E +

(un~-'O)

to the B 2 E + (v' = 0) state. The red-shifted (v' = 0 ~ v" = 1) fluorescence was collected at a photomultiplier tube through an interference filter at 420 nm (35% transmission, fwhm = 10 nm) to minimize interference from scattered light.

NCO molecules were prepared via the rapid reaction (3a)

CN + 0 2 --* NCO + O, (3a)

CN + 0 2 ~ NO + CO, (3b)

in which k 3 = 2.1 × 10 -11 cm 3 molecule -1 s -1 [9], with the major product channel (3a) ( ~ 70%) [10,11]. The concentration of 02 was at least 3.6 × 1016 molecule cm -3 so that in 10/xs reaction (3) went to > 99.9% completion. The A(0,2°,0)-X(0,01,0) transition at 414.947 nm was chosen to monitor the concentration of NCO radicals because of its greater intensity and small lifetime (328 ns) [12]. The red- shifted fluorescence was collected through an interference filter at 439 nm (45% transmission, fwhm = 10 nm).

A delay/pulse generator (Stanford Research Sys- tem, DG535) was used to set the delay between the photolysis and the probe lasers. The concentration of CN radicals at a fixed delay time was determined by sitting at a fixed delay for 30 to 100 laser shots to obtain a single fluorescence point in the decay curve. The reaction period was determined by the delay between the firing of the photolysis and the probe lasers. The lasers were generally operated at a repeti- tion rate 10 Hz.

The precursors of NCO (ICN, BrCN, O~), reactant HzCO and buffer gas (He) were mixed in flexible tubing (length 30 cm) before entering the reactor and slowly flowed (flow velocity 7 - 2 0 cm s -1) such that the reaction zone was replenished with a fresh gas mixture for each photolysis laser pulse.

The pressure of the system was measured with a MKS 122A Baratron gauge. Calibrated Tylan FM360 mass flowmeters were used to measure the flow rates of reactants and buffer gas. The temperature of the reaction was regulated within 5-1 K with an Omega CN9100A temperature controller and measured with a K-type thermocouple placed 5 mm above the detection region.

The concentrations of the reactants were calculated according to

[A] = 9.66 × IO18PFA/F.r T (molecule c m - 3 ) ,

(4)

in which P is the reaction pressure (in Tort), T is the reaction temperature (K), and FA, F T are the flow rates (in STP cm3/s, STP = 273.15 K, 760 Torr) of reactant A and the total reaction mixture, respectively. The initial CN radical concentration produced by photolysis is estimated from

(3)

1 0 . . . . i . . . . i .... i .... i,.

in which X = I or Br; o" is the absorption cross section of I C N at 248 nm (a value 4 × 10 -19 cm 2 was used) [13], or BrCN at 193 nm (3.3 × 10 -19 cm 2 was used) [14]; F is the fluence (photons c m - 2 ) of the dissociation laser.

Gaseous He (99.999%) and 0 2 (99.97%) were used without further purification. I C N (Aldrich, 95%) was degassed before use. BrCN (Aldrich) was puri- fied by f r e e z e - p u m p - t h a w cycles. H2CO monomers were prepared by heating the polymer (parafor- maldehyde) to 350 K, collected at 78 K and diluted in He to form a 0.5% mixture.

3. R e s u l t s

All kinetic measurements were made under pseudo-first-order conditions with [ H 2 C O ] m o r e than 200 times greater than [CN] or [NCO]. The initial concentrations for CN and N C O were less than 4 × 1011 molecules cm -3, to avoid possible radical-radical reactions and to ensure pseudo-first- order conditions. The rate equation is expressed as

- d [ R ] / d t = k I [ R ] = ( k d + k n [ H 2 C O ] ) [ R ] , (6) in which R = CN or NCO, k I (in s - 1 ) is the first- order decay coefficient, k o is the rate coefficient for radical loss due mainly to diffusion and k n is the second-order reaction rate coefficient of interest.

C N + H2CO reaction. The rate coefficient of reaction (1) was measured in the temperature range 2 9 4 - 7 6 9 K. Fig. 1 shows a typical set of CN decay plots whose slopes are used to determine the first- order rate coefficients k 1. From these slopes we derived the second-order rate coefficients, k r At 294 K, k 1 was measured for pressures in the range 5 2 - 2 0 1 Torr of He; we observed no pressure dependence of the rate coefficient (Fig. 2). We obtained a value (1.64 + 0.04) X 10 -11 cm 3 molecule -1 s - l ; the uncertainty represents two standard deviations in the linear least-squares fit. Combining the estimated errors from the measured parameters (pressure, temperature, gas flow rates, slopes of decay plots) with possible systematic errors we estimated an overall uncertainty for k 1 about 15% at the 95% confidence level. Our value is in excellent agreement with that determined by Lin and co-workers [6]: (1.66 + 0.04)

x 10 -11 cm 3 molecule -1 s -1. A u) t- > , i _ i _ J= Z O 0.2 ,,,,i . . . . i . . . . I, .. l l . 1 2 3 4

Y.-W. Chang, N.S. Wang/Chemical Physics 200 (1995) 431-437 433

T i m e ( 1 0 "4 s )

Fig. 1. Decay plots of CN at 769 K and 90 Torr with [H2CO] (1014 molecules cm-3)=0, k l= 1251 s -1 (17); 1.5, k l= 6285 s -1 (A); 2.5, kl = 10185 s -1 (O); 4.3, k I =17679 s -1 (A).

Plots of k I versus [H2CO] at various temperatures are shown in Fig. 3. All of the plots exhibited satisfactory linearity. The experimental conditions and k 1 measurements of this work are summarized in Table 1. Fig. 4 shows the Arrhenius plots o f k 1 from our measurements. A linear fit gives k I -- (6.7

-t- 1.0) X 10 -11 e x p [ ( - 4 1 2 + 2 0 ) / T ] cm 3 mole- . . . . | . . . . i . . . . 2 A o 0 , , , , I , , , , I , , i i 0 0.5 I 1.5 [H2CO ] (10 Is molecules cm "3)

Fig. 2. Plots of pseudo-first-order decay rate of CN, k 1, versus [H2CO] at 294 K and 52 Torr (Q); 100 Torr (A); 155 Torr (n); 201 Tort ( * ).

(4)

434 Y.-W. Chang, N.S. Wang/Chemical Physics 200 (1995) 431-437 3 v 0 , . , , I J , , , I , , , , 0 0.5 I 1.5 [H=CO] (10 Is m o l e c u l e s c m "a)

Fig. 3. Plots of pseudo-first-order decay rate of CN, k 1, versus [H2CO] at 323 K ( 0 ) ; 357 K ( + ); 400 K ( • ); 455 K (O); 526 K (11); 625 K ( v ) ; 769 K ( * ) .

cule -] s - l , the error limits are at the 95% confidence level.

NCO + H2CO reaction. The rate coefficients of

reaction (2) were measured over the temperature range of 294-769 K using CN + 0 2 as a source of NCO. The temporal profile of [NCO] (Fig. 5) shows a slow decay (k I = 200-1200 s -1, which varied with temperature) with no H2CO present. The decay was probably due mainly to diffusion of NCO away from the detection zone. The rate of decay of NCO

A 9 "7,

e

:3 5 O O

E

O 2

6

v . , , I . . . . i . . . . i . . . . I . . . . i . . . . 1.5 2 2.5 3 3.5

1000/T (K 1)

Fig. 4. Arrhenius plots of kt; this work ( 0 ) ; Yu et al. (O).

Table 1

Summary of CN + H2CO reaction data

T Pressure [H2CO] × 10-14 kl × 10 H a

(K) (Torr) (molecules c m - 3) (cm 3 molecule- 1 s - 1) 294 52-201 1.3-12.6 1.645:0.04 b 323 100--105 1.3--14.1 1.945:0.04 357 95--100 1.0--12.1 2.035:0.02 400 95--100 0.9--11.0 2.325:0.04 455 98--104 0.9--10.4 2.74 5:0.02 526 92--98 0.9--9.7 3.115:0.02 625 88--94 0.6--7.0 3.23 5:0.02 769 89--94 0.5--5.7 4.08 5:0.04 a kl(T )= (6.75: 1.0)× 10 -11 e x p [ ( - 4 1 2 + 2 0 ) / T ] cm 3 molecule- 1 s - 1.

b The indicated error limits are 2 o-.

1 0 . . .. i . . .. i . .. .

5'

zo

1

0 0.5 I 1.5

Time ( I 0 "a s )

Fig. 5. Decay plots of NCO at 455 K with [H2CO] (1015 molecules cm - 3 ) = 0, k I = 320 s - 1 ( [] ); 0.2, k I = 828 s - 1 ( O ) ; 0.5, k I = 1396 s -1 (A); 0.9, kl = 2408 s - t (O); 1.1, k I = 3 1 4 8 s -1 ( n ) .

Table 2

Summary of experimental conditions of the reaction NCO + H 2 CO T Pressure [H2CO]× 10-14 k 2 X 1012 a

(K) (Torr) (molecules c m - 3) (cm 3 molecule- 1 s - 1) 294 92-105 1.5-22.6 1.57 5:0.03 b 323 90--100 1.3--20.4 1.63 ± 0.02 357 90--100 1.2--19.5 1.87±0.03 400 90--100 0.9--15.1 2.19±0.05 455 100--110 0.9--12.4 2.51 5:0.08 526 90--100 1.1--15.7 3.065:0.03 625 90--100 1.0--11.6 4.11 5:0.06 769 90--100 0.8--8.9 5.16±0.09 a k2(T ) = 4.62 × 10-17 T1.71 e x p ( 1 9 8 / T ) cm 3 molecule- 1 s - 1. b The indicated error limits are 2 tr.

(5)

Y.-W. Chang, N.S. Wang/Chemical Physics 200 (1995) 431-437 435 5 ~'~ 4 0 3 v ~ 2 1 0 0

... ; ' / 7

p

!'t

0.5 1 1.5 2 2.5 [HzCO ] (10 is m o l e c u l e s c m "-~)

Fig. 6. Plots of pseudo-first-order decay rate of NCO, k l, versus [H2CO] at 294 K (Q); 323 K (~); 357 K ( [] ); 400 K ( • ); 455 K (O); 526 K (n); 625 K (V); 769 K (0).

increased significantly upon addition of H2CO and the coefficients k ~ for the first-order decay rate were calculated from slopes of the semi-log plots. By fitting k I values versus [H2CO], as shown in Fig. 6, we obtained k 2 from the slope. Table 2 summarizes the experimental conditions and results; the error limits represent 2,, of our work on reaction (2). These measurements show that the temperature de-

10 U) "7

i

o l 1 J i l l | , . I . . . . i . . . . i . . . . i , . , 1.5 2 2.5 3 3.5 4

10o0/I" (K

Fig. 7. Arrhenius plots of k 2.

pendence of k 2 deviates from the Arrhenius form: k 2 = 4.62 × 10-17T 1"71 e x p ( 1 9 8 / T ) cm 3 molecule-1 s-1, as plotted in Fig. 7.

4. D i s c u s s i o n

C N + H2CO reaction. Yu et al. [6] recently pub- lished results of k 1 measurements (also shown in Fig. 4). They obtained a non-Arrhenius temperature dependence of k 1 between 297 and 673 K: k I = 2.82 X 10-19T 2'72 e x p ( 7 1 8 / T ) cm 3 molecule -1 s -1. They also performed transition state theoretical calculations on reaction (1). The calculated and measured rate coefficients agreed reasonably well when a simple H-abstraction mechanism with loose transition state model was employed. Their measurements and this work are in good agreement at room temperature and at about 600 K but deviate approximately 1 0 - 1 4 % in between, which is within the error limits of the measurements. However, the deviation is large when extrapolated to a flame temperature. Additional measurements at higher temperatures are needed before k I is applied to combustion models.

Yu et al. found that k I varied with the energy of the dissociation laser ( ~ 2 0 - 1 0 5 m J / p u l s e ) : k 1 decreased with increasing laser energy. They attributed this dependence of k I to the photodissociation of H2CO and determined k 1 using a small laser energy ( < 30 m J / p u l s e ) to minimize this problem. In this work, the rate coefficients were determined at an energy 0 . 5 5 - 1 . 2 0 mJ of the dissociation laser (beam diameter of 5 mm) and we detected no dependence of k 1 on laser energy. The absorption cross section of formaldehyde at 248 nm is ~ 1.0 × 10 -21 cm 2 molecule -1 (296 K) [15] and that at 193 nm is expected to be smaller than that at 248 nm. The loss of reactant due to photodissociation was negligible ( < 0.0002%) under our experimental conditions. In a data evaluation Tsang [7] estimated k I = 7 × 10 -11 cm 3 molecule -1 s -1 based on the measured rate coefficient of the similar H-atom abstraction reaction CN + C3H 8 (k = 3.3 × 10-13T °'56 e x p ( 6 4 9 / T ) cm 3 molecule- 1 s - 1 ) [ 8 ] . His estimate is about four times

as large as our result.

The reactions of CN with unsaturated hydrocarbons C2H 2 and C2H 4 are believed to proceed by a

(6)

436 Y.-W. Chang, N.S. Wang / Chemical Physics 200 (1995) 431-437

mechanism of addition followed by H-atom elimination. The rate coefficients of those reactions feature nearly gas kinetic values (2.5 × 10 -1° cm 3 molecule-I s-1 at room temperature) and a slight nega- tive temperature dependence ( E a = - 0 . 3 5 kcal mo1-1) [16]. However, CN reacts with H2CO about 1 / 1 5 times as rapid as with C2H 2 and C2H 4 at room temperature and with an activation energy near 1 kcal mo1-1. Hence reaction (1) may occur via a simple H-abstraction mechanism analogous to the reaction CN + C 2 H 6. The rate coefficients of these two reactions are similar in both magnitude and temperature dependence [17]. Formation of an adduct of CN added to the "rr system of formaldehyde is unlikely as the occupation of the anti-bonding 2b 1 orbital by the unpaired electron of CN is energeti- cally favored with a structure of the H - C - O angle near 90 ° that deviates significantly from its equilib- rium structure. Lack of a pressure dependence of k 1 may further indicate a direct H-abstraction.

NCO + H2CO reaction. The upward curvature on

the Arrhenius plots of k 2 indicates that the activation energy of reaction (2) increases with increasing temperature. This phenomenon could be explained according to transition state theory with a loose transition structure [18]. The curved Arrhenius plots might also arise from quantum-mechanical tunneling.

We know of no previous investigation of the kinetics of reaction (2). Only an estimate k 2 -- 1 ×

10 -11 cm 3 molecule -1 s -1 was reported by Tsang [7], which is about ten times our value. The kinetics of reactions of NCO with hydrocarbons are investigated by several researchers [19-21]. Perry reported that the rate coefficient of the reaction NCO + C2H 4 depended on pressure and decreased with increasing temperature [19]. Park and Hershberger observed a slight amount of HNCO produced from the reaction of NCO with C2H 4. An addition-elimination mechanism is evident for the reaction NCO + C2H 4. In contrast, our measurements suggest that NCO radicals may not add to the double bond of H2CO. Instead, reaction (2) occurs mainly via a simple H-abstraction path similar to reaction (1) with the same rationale.

We endeavored to evaluate and to eliminate all possible interfering reactions to ensure the accuracy of our kinetic measurements. The NCO source, reaction (3), also produces O, NO, and CO, with k 3 b / k 3 a

= ( 0 . 3 - 0.4) [10,11]. Oxygen atoms and NO are known to react rapidly with NCO:

NCO + O -~ products, (7)

NCO + NO ~ products, (8)

in which k 7 = 1.7 × 10-11 c m 3 molecule- 1 s - 1 [22], and k a = 3.4 × 10 -11 cm 3 molecule-1 s-1 [7]. Un- der the experimental conditions [NCO] 0 = 4 X 1011 molecule cm -3, the concentrations (in 1011 molecule c m - 3 ) of O, NO, and CO are approximately 4, 1, and 1, respectively. Hence a first-order loss of NCO < 7 s -1 would be due to reactions (7) and

(8),

which is negligible. Oxygen atoms produced in reaction (3a) react with H2CO, k(300 K) = 1.7 × 1 0 - 1 3 cm 3 molecule-1 s-1 [21]. Possible interference from this reaction was negligible under pseudo-first-order conditions, [H2CO] >> [O] = [NCO] 0.

NCO radicals produced from reaction (3) are known to be vibrationally excited [23]. A non-Boltz- mann distribution of internal energy of the NCO reactant might introduce errors into the kinetic measurements because of slow relaxation to populate the monitored level and to the effects of internal energy on the reactivity of the radical toward H2CO. No such error was expected in this work as all measurements were undertaken under pressures at least 90 Torr and relaxation periods greater than 10 /zs to ensure a thermally equilibrated environment.

Photodissociation of formaldehyde produces reac- tive species,

HECO + h v ( 2 4 8 , 193 nm) ~ H 2 + CO, (9a) H2CO + h v ( 2 4 8 , 193 nm) ~ H + HCO, (9b) that may react with CN and NCO radicals. At the largest concentration of H2CO employed in this work, the concentrations of photodissociation products were less than 2 × 101° molecule cm -3. Thus they cannot cause interference with the kinetic measurements.

In summary, we determined the rate coefficients of reactions of CN and NCO with H2CO at 294-769 K using laser photolysis-laser induced fluorescence method. The measured values of k I are in reasonable agreement with the only published results. No kinetic measurement of k: is previously reported. Our investigation of the title reactions provide important information to improve understanding of the combustion chemistry of nitrogen compounds.

(7)

Y.-W. Chang, N.S. Wang/Chemical Physics 200 (1995) 431-437 437

Acknowledgements

We thank the National Science Council of ROC (Contract No. NSC 83-0208-M-009-042) for support.

References

[1] C.F. Melius, Proc. 25th JANNAF Combustion Meeting, CPIA Publication 498 2 (1988) 155.

[2] J.A. Miller and C.T. Bowman, Progr. Energy Combust. Sci. 15 (1989) 287.

[3] A.N. Hayhurst and A.D. Lawrence, Progr. Energy Combust. Sci. 18 (1992) 529.

[4] R.A. Perry and D.L. Siebers, Nature 324 (1986) 657. [5] J.A. Miller and C.T. Bowman, Intern. J. Chem. Kinet. 23

(1991) 289.

[6] T. Yu, D i . Yang and M.C. Lin, Intern. J. Chem. Kinet. 25 (1993) 1053.

[7] W. Tsang, J. Phys. Chem. Ref. Data 31 (1992) 753. [8] W.P. Hess, J.L. Durant and F.P. Tully, J. Phys. Chem. 93

(1989) 6402.

[9] Y.Y. You and N.S. Wang, J. Chin. Chem. Soc. 40 (1993) 337.

[10] W.F. Cooper, J. Park and J.F. Hershberger, J. Phys. Chem. 97 (1993) 3283.

[11] F. Mohammad, V.R. Morris, W.H. Fink and W.M. Jackson, J. Phys. Chem. 97 (1993) 11590.

[12] T.R. Charlton, T. Okamura and B.A. Thursh, Chem. Phys. Letters 89 (1982) 98.

[13] K.E. Holdy, L.C. Klotz and K.R. Wilson, J. Phys. Chem. 52 (1970) 4588.

[14] J.A. Russell, J.B. Halpem, W.M. Jackson and I.A. Mclaren, J. Phys. Chem. 91 (1987) 3248.

[15] J.D. Rogers, J. Phys. Chem. 94 (1990) 4011.

[16] D.L. Yang, T. Yu, N.S. Wang and M.C. Lin, Chem. Phys. 160 (1992) 317.

[17] D.L. Yang, T. Yu, N.S. Wang and M.C. Lin, Chem. Phys. 160 (1992) 307.

[18] I.W.M. Smith, Kinetics and dynamics of elementary gas reactions (Butterworths, London, 1980) pp. 198-205. [19] R.A. Perry, 25th Symp. (Intern.) Combust. (1986) p. 913. [20] S. Wategaonkar and D.W. Setser, J. Phys. Chem. 97 (1993)

10028.

[21] J. Park and J.F. Hershberger, Chem. Phys. Letters 218 (1994) 537.

[22] D.L. Baulch, C.J. Cobos, R.A. Cox, C. Esser, P. Frank, Th. Just, J.A. Kerr, M.J. Pilling, J. Troe, R.W. Walker and J. Wamatz, J. Phys. Chem. Ref. Data 21 (1992) 411. [23] D.G. Sauder, D. Patel-Misra and P.J. Dagdigian, J. Chem.