J. Phys. Chem. 1993,97, 7234-7238
7234
Quenching and Relaxation of Vibrational Levels of NH/ND(alA, v) B. Bohn and F. Stuhl' Physikalische Chemie I, Ruhr-Uniuersitiit Bochum, 0-44780 Bochum, Germany Receiued: February 25, 1993; In Final Form: April 27, I993
The kinetics of the removal of vibrational levels of N H and N D in the metastable (alA) state was investigated for collisions with the parent molecules HN3, HNCO, DN3, and DNCO and with Nz and 0 2 . The parent molecules exhibit neither isotope effects nor a dependence of the removal rates on the vibrational levels. Opposite isotope effects were observed for Nz and 0 2 ; the rates become larger for Nz and smaller for 0 2 upon deuteration. For both 0 2 and Nz, the rate constants increase with increasing vibrational excitation. In the case of Nz, levels with u > 0 are barely electronically quenched but efficiently relaxed most likely in Au = -1 steps to lower vibrational levels. The marked decrease in the quenching efficiency upon vibrational excitation is in qualitative agreement with efficient mixing of the N-H stretch with the other vibrational modes in the N r N H complex and the Landau-Zener description of diabatic crossing of potential energy surfaces. Rate constants are estimated for the electronic quenching and vibrational relaxation processes.
Introduction Metastable NH(a1A) radicals are commonly generated in the UV photolyses of hydrazoic acid and isocyanic acid. For the past 15 years, the determination of the degree of relative vibrational excitation of the radicals has been the aim of a number of experimental studiesl-lo yielding various results. Recently we have reported on the vibrational distributions of NH/ND(a,u) in the photolyses of HN3/DN3 and HNCO/DNCO,lO which are in good agreement with those of previous indirect9JlJ2 and direct9 studies and a very recent investigation of the wavelength dependence of the NH(a) u = 0 to u = 1 and u = 2 population ratio by Hawley et al." Such knowledge of the vibrational excitation is necessary for the interpretation of kinetic experiments. Nonexponential decays of NH(a,u=O) could not be explained? before it was apparent that vibrationally excited NH(a,u>O) is formed3 and, in the presence of Nz, NH(a) in u = 1 is relaxed to u = 0.6~'~In the quenching of NH(a) by 0 2 , the product Oz(blZ,+) has been observed.ls Up to now it is not evident which role vibrationally excited radicals play in this endothermic energy transfer process. If vibrational energy is of importance, isotope effects can be expected for NH/ND. However, little is known of the kinetics of ND(a,u). In view of the new data on the vibrational populations of metastable NH/ND(a,u), we have initiated this study on the kinetics of vibrational levels of NH/ND(a,u). We will show that reliable population data and precise kinetics can, in some cases, lead to information on both vibrational relaxation and electronic quenching. A preferably relaxing gas can be desirable to modify the NH(a,u) vibrational population and thus change the kinetics, if vibrational excitation is involved. Experimental Section Metastable NH and ND radicals in the (alA,u) state were generated by excimer laser light pulses in the 248- and 193-nm photolyses of HN3 and DN3 and in the 193-nm photolyses of HNCO and DNCO. Most experiments were performed using the 248-nm photolysis. The radicals were detected by dye laser induced fluorescence (LIF) on the P(2) or P(3) lines of the (0-0) band of the the (clIl-alA) transition for NH/ND(a,u=O), the (0-1) band for NH/ND(a,u=l) and the (1-2) band for NH/ ND(a,u=2). The experimental system has been described previously10 and only the relevant features will be mentioned here. Moreover, diffusion effects have been considered in some detail to judge the accuracy of the data. 0022-3654/93/2097-7234S04.00/0
For the detection, the dye laser was tuned onto the maximum of an excitation line. Time profiles of relative NH/ND(a,u) concentrations were usually generated by LIF intensity measurements at about 20 different delay times between photolysis and analysis laser. Different from our previous experiments,1° saturation effects are not important. First, the signals from 10 dye laser shots at 10 Hz were averaged in a gated integrator to be fed to a PC. About 60 of these averages were then used to form a mean value of the LIF signal at a chosen delay. Care was taken to keep the intensities of the excimer photolysis laser and the dye analysis laser constant during a run (about 40 min). This was checked by reversing the order of the delays after 20 min until the original delay was reached. The time profiles were evaluated by a homemade program (CURFIT) which takes Poisson statistics into account. All kinetic data were determined in the presence of 3 kPa of He. As will be described below, this pressure sufficiently reduces diffusion effects. The addition of He furthermore relaxes rotational but not vibrational excitation. The following maximum pressures were used in the kinetic experiments: Nz,700 Pa; 02,250 Pa; HN3/DN3,1.2 Pa; HNCO/ DNCO, 1.2 Pa. For the data obtained in the presence of N2 and 0 2 , the typical pressure of HN3 and DN3 was 0.05 Pa. The stated (Messer-Griesheim) purities of the gases were Ar, 99.998%; He, 99.996%;02,99.995%; N2, 99.999%. Hydrazoic acid and isocyanic acid were generated and handled as described previously.lo The isotope purities of the deuterated species usually was 70-80% but sometimes lower. Results Diffusion Losses. A number of runs and some simulations were performed to study the contribution of diffusion to the measured decays of NH(a) in order to learn how to minimize diffusion from and to the observed volume and hence to increase the accuracy of the decay time measurements. The stable resonator of the excimer laser generates light pulses travelling in the x direction, which are not significantly attenuated by the gases of the photolysis system. Its rectangular cross section (in they and z directions) is about 2.3 cm X 1 cm. The dye laser beam has a circular cross section of 1-mm diameter and is aimed in the y direction to intersect the excimer laser beam perpendicularly along the long side of its cross section. The volume formed by the intersection represents the observed photolysis volume. The net flux of NH/ND(a,u) into and out of this volume determines the diffusion loss rate. It should be noted that by the use of lenses, the geometry of the laser beams was changed from this original setup for the kinetic experiments. 0 1993 American Chemical Society
Vibrational Levels of NH/ND(alA,u)
t .
0
The Journal of Physical Chemistry, Vol. 97,No. 28, 1993 7235
w
rlcm I
0.2
1 .
I
.
I
.
0.4 0.6 timelms
1
0.8
.
1
1
K m 1. Decay curves of NH(a,u=l) radicals in a semilogarithmic plot. The radicals were generated by photolysing 0.07 Pa of HN3 in 1 kPa of He. The detection was performed on the P(2) line of the NH(CL1) band at 363.6 nm. Curve (a) is the observed decay with the original laser beam profiles, for (b) the cross section of the excimer laser beam was half of the original size, (c) representsa pure exponential decay as derived from (a) and (b) for the absence of diffusion. The inserts represent the reconstructed laser profiles in z direction. The net flux of NH/ND(a,u) was modeled for the following assumptions and simplifications: (i) The net fluxes in the x and y directions were considered to be negligible and only the flux in the z direction counts. (ii) The excimer laser fluence in the z direction is of Gaussian shape and the photolysis is a linear process. (iii) The diffusion of NH(a,u) in the z direction can be described by Fick's second law. We obtain for the number density of N H or ND(a,u=O, 1, or 2), N(z,t), the following expression:
In this equation, D is the diffusion coefficient, a a parameter characteristic for the Gauss distribution 2-( = halfwidth), and T the lifetime, which would be measured in the complete absence of diffusion. With the LIF intensity signal, IF(t), proportional to the total number the radicals in the observed volume, we obtain
which can be solved numerically. The origin of the Cartesian coordinate system is the center of the photolysis volume. The photolysis volume extend from -a to a in the x, from -b to b in they, and from -c to c in the z dimension. In this original geometrical setup, the dye laser cross section is so small that it monitors mainly the maximum of the Gauss distribution on the z axis. We therefore adopt for a moment the approximation z = 0 and obtain from eq 1
zF(t)a e - r / T / d i T i i G
0.4 0.8 p(HNCO/DNCO)/Pa
1
(3) Figure 1 represent two measured decays of NH(a,u= l ) , (a) with the original beam arrangement and (b) with the excimer laser beam cross section approximately halfed by using a lens. Fitting the data of curves (a) and (b) to eq 3 results in T = 480 ps for both curves and D a = 470 and 1420 s-1, respectively. The straight line in this figure represents the pure exponential decay for the lifetime T = 480 ps. The curves of Figure 1 show that it will not always be easy to distinguish pure exponential decays from nonexponential ones unless the precision of the data is sufficiently high. From the knowledge of D, which can be estimated from D (NH3 or Nz or 0 in 10s Pa of He) = 0.6 to 0.7 cm2 s-1 l 6 and
1.2
Figure 2. Decay rates of NH/ND(a,u=Oand 1) as a function of prcssure of HNCO/DNCO. The curves for u = 1 are displaced by +5000 s-1.
Dp = constant, a can be determined to reconstruct the laser profiles. Good agreement with the actual beam dimensions was obtained. Simulations using eq 2 show that diffusioncan be best minimized by increased pressures and expansion of the excimer laser beam. Expansion of the dye laser beam is less effective as long as the dye laser beam dimensions are smaller than those of the excimer laser beam. Equation 3 can be simplified for 4Dat < 1 to give
zF
e4~'+2Du)r
(4) In this approximation, one recognizes that a decay rate, T,~-* = 4 2Da, is defined, which can be determined experimentally. The usual plots of ~ ~ = ~TO-1 -+ k[R] 1 then result in values of rate constants, k,for reactant concentrations [R] in excess over the concentration of the transient species. This approximation is valid for large total pressures (small values of D), short times and small values of a. Furthermore upon addition of R,the change of D must be small. For the value of a = 6.9 cm-2 as derived from curve (a) in Figure 1 and D (NH(a) in 3 kPa of He as used in the final runs) = 25 cm2 s-l, eq 4 is useful for t < 1.5 ms and 2Da = 300 s-l. To be on the safe side, we have additionally expanded the dye laser beam which slightly decreases the diffusion effect. The main advantage of this expansion, however, is less saturation in a larger observed volume and therefore more efficient use of the dye laser photons and furthermore less chance of hitting the flank of the excimer laser profile with the confined dye laser beam. The diffusion coefficient D does not only depend on pressure but also on the nature of the buffer gas. The value of D for Ar is about 3 times smaller than that for He.16 Ar, however, was not used, because the final beam arrangement and 3 kPa Ar gave intercepts, TO-^, which were between 700 and 1000 9-1 for NH(a,u=l and 2 ) but close to zero for NH(a,u=O). On the other hand, experiments in 3 kPa of He gave intercepts consistently between 200 and 300 s-I for all vibrational levels. Presumably, the rate constant for the relaxation of u = 1 and 2 by Ar is not greater than 1.4 X 1 0 - l s cm3 s-l. We hence suspect that large amounts of Ar can contribute to somevibrational relaxation which is to be avoided for the bath gas in the present study. The constant intercept for He is due to diffusion, quenching by He and impurities, and radiation. Rate Constants. All decays were found to be exponential with the exception of those of NH/ND(a,u=O) in the presence of Nz. Rate constants of the removal of NH(a,u) and ND(a,u) radicals at room temperature were determined from their decay rates, ~ ~ ~ in~ the - 1 presence , of excess of the parent gases HN3, DN3, HNCO, DNCO, and 0 2 and N2. Examples of plots T ~ vs~ pressure of colliding gas are shown in Figure 2 for NH/ND(a,u=O and 1 ) radicals in the presence of HNCO/DNCO. The slopes determine the values of the rate constants, k, which are
+
-
~
7236 The Journal of Physical Chemistry, VoI. 97, No. 28. I993
Bohn and Stuhl
TABLE I: Measured Effective Rate Constants for the
Removal of NH/ND(a,v) from the 248-nm Photolysis of HN3 and DNJ in 3 kPa of He and at T = 300 f 3 K collider N2
0.81 f 0.076
0 2
HN3 HNCOd
0.49 f 0.04 1210 i 90 546f39
N2
1.09 f 0.096
0 2
0.39 f 0.05 1220 f 80 531 f 44
DN3
v=l
v=2
2.87 f 0.12 2.10 f O.2lC 0.80 f 0.07 1090f 110 559 f 30
2.98 f 0.20
v=o
NH(a,v)
ND(a,u)
3.10 f 0.22 2.75 f 0.26c 0.49 f 0.06 1120 f 100 547 f 45
1.44 0.25 1010f 210
3.89 f 0.31 0.83 i 0.07 1080 f 180
DNCOd 0 Error limits represent 3u. Only exponential part of decay curves taken intoaccount. C Valuedetermined in the 193-nmphotolysis. Value determined in the 193-nm photolysis of HNCO/DNCO. given in Table I. The measurements in the presence of 0 2 were more difficult to perform than those with N2, because 02 is an efficient quencher of the excited NH(c) state,” which limited the useful pressure for this relatively slow quencher. Figure 3 presents examples of the nonexponential NH(a,u=O) time profiles in a linear plot for two different N2 pressures. At long times, the radicals are found to disappear exponentially. The correspondingfunctions are displayed in Figure 3 to illustrate the deviations from the exponential dependence at short times. Decay rates and hence rate constants were determined from the curve at long times. Because we suspected feeding of u = 1 from upper vibrational levels, we have additionally studied the kinetics ofNH/ND(a,u=l) in the 193-nmphotolysesofHN3/DN~,which yield vibrational distributions altered from those at 248 nm.Io The result are significantly different rate constants as displayed in Table I. Thus, in a strict sense, the rate data for NH/ND(a,u= 1 and 2) in Table I are ”effective” constants at least for the removal in the presence of N2. They will be evaluated and discussed later in detail. The rate constants for the removal of NH/ND(a,u) by He is estimated to be smaller than 3.5 X 10-16cm3 s-l for u = 0, 1, and 2. These upper limits were obtained from extrapolation of the decay rates to zero pressure of the parent molecules. Discussion NH/ND(a,v) + HNo/DNs and HNCO/DNCO. The kinetics of most of the investigated NH/ND(a,u) levels in collisions with a number of simple molecules has been studied previously and a comparison with these data is possible in several cases. For collisions with the parent molecules, rate constants have been previously reported for NH(a,u=O,),15J8-23 NH(a,u= 1,),6 NH(a,u=2,)6 + HN3; ND(a,u=O,) + DN3;g and NH(a,u=O,) + HNC0.24a2SIn general, there is good agreement among all values with the exception of all literature values for NH(a,u=O) + HNC0.24.25 The reasons for these discrepancies are not evident but might be incorrect pressures of the parent gas, which is difficult to synthesize and handle. To our knowledge there are no values available on NH(a,u=l) + HNCO, ND(a,u=l and 2) + DN3, and ND(a,u=O and 1) + DNCO. The reaction Of NH(a) with HN3 is reported to form N&) + NH2(% and A).I The excited NH2(A) products have been previously used to study the kinetics of NH(a).1J8J In collisions with H-NCO, the products are assumed to be similar?’but excited NHz(A) cannot be formed, because of energy reasons. Within the error limits, the data of Table I clearly show for the parent molecules that (i) the rate constants for removing the different vibrational levels are constant for all investigated levels (ii) no isotope effects are observed upon changing the isotope purity.
61.2 iijj 0.8
c\
5
-
2 0.4 0
0
0.1
0.2
0.3
0.4
0.5
time/ms Figure 3. Time profiles of the relative NH(a,v=O) LIF intensity for two different pressures of N2. Both curves are normalized for the same intensity at time zero. The parts of the curves at long times were fitted to exponential functions which are extended up to t = 0. The intercepts of these functions on the ordinate is about 1.6 times larger than the measured initial intensity. The radicals were generated in the 248-nm photolysis of HN3.
We do not expect a large temperature dependence of the rate constant and, further, suppose that all vibrational states are removed in reactions and do not vibrationally relax. NH/ND(a,v) + 0 2 . The quenching of NH(a,u) by 0 2 has been studied by some research groups and previous work has been discussed in refs 12 and 15. No data have been reported before for ND(a,u) + 02. The data obtained by both the phosphorescence and LIF of NH(a) agree well. Although the value deduced from the phosphorescencemethod is significantly larger [(6.2 f 0.8) X cm3s-~]*~ than the value given in this work for NH(a,u=O), it agrees with the data of this work. Modelling of the NH(a+X) phosphorescence spectrum shows that most of the emission from vibrationally excited NH(a,u) falls into the observed bandwidth of the previously used monochromator. In Table I, the rate constantsfor 02increase markedly with vibrational excitation. Weighting the phosphorescence intensitieswith the relativeefficiency of receiving phosphorescence from the different vibrationally excited states and the relative population of the states yields an effective rate constant close to 6X cm3 s-l as observed by phosphorescence. The data for 02in Table I exhibit kinetic isotope effects, which increase with vibrational excitation of NH/ND(a,u). The rate constants for the deuterated species are smaller. Possible paths of the process are NH(a,u)
-
+ 0,
NH(X,u)
+ 02(X)
(Rl)
NH(X,u)
+ 02(a)
(R2)
NH(X,u)
+ 02(b)
(R3)
-NO+OH
(R4)
-HNO+O
(W
The participation in the NH(a) removal has been reported to be less than 40% for the reactive channels (R4)-(R6).4 Energy resonance processes giving 02(a1A,u)2Sand 02(b1Z,+,u=O)ls have been proposed previously. The generation of Oz(a’A,u) has been suggested from indirect experiments when detecting NH(X,u).Z5 Time-resolved formation of 02(b) has been observed by emission during the decay of NH(a). Although the accuracy of the rate of growth of Oz(b) and the intensity ratios prevented a firm conclusion, support was brought forward in favor of an efficient
Vibrational Levels of NH/ND(alA,u)
The Journal of Physical Chemistry, Vol. 97, No. 28, 1993 7237
process (R3).’5 Reaction (R3) is endothermic for the reactant NH(a,u=O) by 433 cm-1. An activation energy of 720 cm-1 has been measured recently.26 Qualitatively the trends of the isotope effects in Table I can hence be understood by the relative energetic positions of the vibrational levels in NH/ND(a,u). Work is in progress in our laboratory to study the contribution of vibrational levels in the energy-transfer process (R3). ThedistributionofthevibrationallevelsoftheNH(X,u) product has been previously reported to be colder than that generated in collisions with Xe.l2 There are indications that quenching of NH(a,u) by Xe is a vibrationally adiabatic process.8*9J2 It can hence be concluded that some of the vibration of NH(a,u) gets dissipated during the collision with 0 2 . Processes which have to be considered are vibrational relaxation within NH(a,u) and/or simultaneous change of vibration and electronic state during a collision with 0 2 . Since no deviation from exponential decays is observed in the presence of 02, vibrational relaxation appears to be a minor process. NH/ND(a,v) + N2. The kinetics of the system NH/ND(a,u) + N2 seems to be more complex as indicated by the nonexponential decays of NH/ND(a,u=O) and the different rate data for NH/ND(a,u=l) which depend on the photolysis wavelength (Table I). It is so much more surprising, that the rate data of Table I are in agreement with all previously measured rate constants.4.6~1’,12,15.24.26 No data are available on ND(a,u= 1 and 2) + N2. Among the previously measured values that reported by Freitag et al.15 appears to be too low but can be explained by the phcsphorescence monitoringmethod. This technique monitors all vibrational levels but with some preference of u = 0 and thus some relaxational effect cannot be avoided. How relaxation lengthensthedecayofNH(apr0) at shortreaction timesisclearly demonstrated in Figure 3. Temperature-dependent rate constants have been measured for NH(a,u=08*26and la) N2 and ND(a,u=O) + N2.8 Within theerrorlimits, theactivationemgies, E,, are all found to be the same (=5 W mol-l). The calculated value of Ea is about 2.5 times larger.27 It should be noted for N2 that an isotope effect has been observed which gives larger rate constants for ND(a,u=0),8 opposite that found for 0 2 . In this work, this trend is additionally confirmed for u = 1 and 2. Efficient vibrational relaxation of NH(a,u) was first indicated by the nonexponential decays in the presence of N2 reported by Rohrer14 and Hack and W i l m ~ .Rohrer ~ demonstrated that NH(a,u=O) was additionally generated while u = 1 was rapidly removed. On the other hand, Ar was found to remove both vibrational levels at the same rateand didnot givenonexponential decays.14 The only possible product of the quenching of NH(a) by N2 is NH(X). Adams and Pasternacklz and Hack and Rathmanns have observed that NH(X) is formed only in u = 0 which is produced with a rate similar to that of the removal of NH(a,u=O). These facts are consistent with rapid relaxation of NH(a,u) to NH(a,u=O) before it is quenched to NH(X,u=O). In the following we will provide further evidence for this process. The only source for the additional feeding of the NH(a,u=O) level at short times we can think of is relaxation from upper vibrational levels. The exponential curves displayed in Figure 3, when extrapolated to time zero, represent a measure of the sum of the original NH(a,u=O) and the relaxed population. Hence the ratio of the intercepts of the exponential and experimental curves give the sum of the populations originally formed in u = 0 and subsequently relaxed to u = 0 relative to that originally formed in u = 0. Evaluations of plots like those of Figure 3 gives ratios ranging from 1.53 to 1.73 in the photolysis of HN3 and from 1.65 to 1.85 in the photolysis of DN3 at 248 nm. These values were found to be independent of the N2 pressure. The corresponding previously measured population ratios are 1.63 and 1.65 for HN3 andDN3, respectively.1° Also thesedata support a quantitative relaxation process. On the basis of the original relative populationsin the vibrational
100
1
T
I
p(N2)/1O2 Pa Figure 4. Plots of k:[N2] for NH and ND as a function of N2 pressure.
levels and the “effective” rate constants of Table I, we will now develop a simplified model for vibrational relaxation of NH/ ND(a,u) by N2. As mentioned before we adopt that He and HN3/DN3 does not contribute to relaxation. Furthermore the different rate constants for NH/ND(ap= 1) at the two photolysis wavelengths (yielding different vibrational distributions) imply that at least some of the relaxation by N2 occurs via steps Au = -1. Therefore we propose the following mechanism is valid for NH(a,u):
-
+
products
Vibrational states higher than u = 2 are neglected here because of their small populations, The same mechanism can be written for ND(a,u) as well. The rates of the relevant processesare given next to the arrows. k,” is the rate constant for relaxation of vibrational level u to u - 1 by N2; &: is the rate constant for quenching of the vibrational level u, respectively. Q is the quenching partner N2 or HN~/DNJ. We start with describing the time &g”x of NH(a,u=O) knowing two additional facts: (i) NH(a,u= 1)dtcaysexponcntiallyac~ordiagto thedecay rate k‘, = ‘“4 k y [ H N 3 ] with ’@a being the “effective” removal COMtaat d u = 1 for the photolysis wavelongth X as given in t b third column of Tabk I. (ii) The original population ratio of u = 1 and u 0, &, depends on the photolysis wavelength A. Soking t L agpropriate differential equation for NH(ap=O) gives
+
[NH(ap=O)], =
+
In eq 5, ko = kp[N2] k y [ H N 3 ] . Fits of the time profiles of NH(a,u=O) toeq 5 using CURFIT resulted in values of ko and e [ N 2 ] . It should be noted that the experiments were performed for h = 248 nm only, because of too little variation of &between h = 193 and 248 nm. Thevalues of hobtained from the fits were the same as those obtained by evaluating the NH/ND(a,u=O) decays at long times. Plots of ky[N2] vs pressure of N2 are shown in Figure 4. They result in straight lines and hence relaxation rate constants, ky,which are given in Table I1 in the third line. The extrapolations of both lines go through the origin indicating that the assumption on negligible relaxation by He is justified. We now consider the time dependenceof NH(a,u= 1) and accept the following two facts: (i) thcdecaysof NH(a,u=2) areobserved
7238 The Journal of Physical Chemistry, Vol. 97, No. 28, 199'3
TABLE II: Rate Constants for the Quenching and Relaxation of NH/ND(a, F1,Z) by NZ rate constant
kp'+ k:
ky
NH
ND
3.0 f 0.3" 2.7 & O.gb 3.3 f 0.3b 0.3 f 0.4
4.9 f 0.50 4.8 & 1.7b 4.8 f 0.6* 0.0 f 0.8
E Obtained from the nonexponential decay of NH/ND(a,u=O), eq 5. Calculated by using eq 7.
to be exponential; (ii) the population ratio of u = 2 and u = 1, &, is known.10 Solving the appropriate differential equation results in a time dependence for [NH(a,u=l)], of the same form as eq 5 but each vibrational quantum number is increased by 1. Additionally, now kl = ( k y + k$)[N,l + krN3[HN31. The decay rates for u = 1 and u = 2 are nearly similar in the presence of Nz and hence kl and k)2 = ' k : q ~ ~+] kyN3[HN3]are not very different. Moreover, tr#~k,,k:[N,] < 1. With appropriate approximations, one obtains [NH(a,u=l)], = [NH(a,u=l)], -0 e-(k14i1k:[NZ1)f
(6)
Introducing the expression for kl into eq 6, the effective rate constant at the photolysis wavelength X is given by
(7) k$ is known from the time profiles of NH(a,u=O) and c$;,l from our previous study.10 The measurements of the effectiveremoval rate constants of u = 1 at the two wavelengths of 193 and 248 nm (Table I) allows us to determine k y and k;. The values of kf'" + k$, k?, and k? are listed in Table 11. We note that the values of k: are very similar to those of k: showing that the relaxation rate constant is not strongly dependent on the vibrational level. One furthermore recognizes that ND(a,u) is more rapidly relaxed than NH(a,u) probably because of the smaller energy spacings. The values for k y are small and probably negligible, and the removal of the excited vibrational levels mainly occurs by relaxation. IR dissociation studies of HN3 show in the range of the 5 q to 6vl overtoneN-H pumping that ground state NH(XQ-) radicals are formed due to extensive mixing of vibrational states.28 At higher energies, IR photodissociation spectra show a transition from discrete to continuous structure in the range of the sixth to seventh overtone of the N-H stretch vibration.29 This can be understood by an increasing coupling of the various vibrational modes and the increasing rate of dissociation. In this energy range (17 670-20 070 cm-l), the dissociation product changes completely fromNH(X3Z-) toNH(a1A).29 In the reverse process, the colliding NH(a) + N2 pair has to overcome a small barrier. The rate constant exhibitsa relatively low A factor and the collision complex decays mainly back to the reactants. The formation of NH(X) is spin forbidden and is due to spin-orbit interaction of the lowest singlet and triplet potential surfaces of hydrazoic acid.30 Ab initio calculations have shown that the lowest p i n t of the seam defining the locus of the crossing of the triplet and singlet surfaces lies about 1000 cm-I below the top of the barrier encountered in the collision NH(a,u=O) + N2.27 The formation of NH(X), and thus quenching in our case, is possible only close
Bohn and Stuhl to the seam for a complex not too rapidly traversing the interaction region according to the Landau-Zener approach.30 Vibrationally excited NH(a,u) have to overcume a similar barrier as D 3 0 radicals11 and form an internally hotter complex. By efficient vibrational coupling via anharmonic mixing of the N-H stretch with other vibrational motions in the intermediate HN3 complex, it appears possible to channel the original vibrational energy of NH into the HN-N2 dissociation coordinate and hence the intersection region of the surface will be passed faster minimizing singlet-triplet interaction according to the generalized Massey parameter.30 In other words, the distance of the lowest point of the seam from the top of the barrier becomes larger for vibrationally excited NH(a,u) and hence the interaction time in the region of the minimum singlet-triplet crossing becomes shorter. This qualitatively explains why the spin forbidden quenching process proceeds less likely for vibrationally excited radicals. A similar argument has been ramtly used to explain the formation of NH(a,u) on the HN3(X1A') surface in the reaction of H + N3 which forms even hotter complexes.31 Acknowledgment. Financial support by the Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie is gratefully acknowledged. References and Notes (1) Baronavski, A. P.; Miller, R. G.;McDonald, J. R. Chem. Phys. 1978, 30, 119. (2) Dekoven, B. M.; Baronavski, A. P. Chem. Phys. Le??.1982,86,392. (3) Rohrer, F.; Stuhl, F. J. Chem. Phys. 1988,88, 4788. (4) Hack, W.; Wilms, A. J. Chem. Phys. 1989,93, 3540. (5) Nelson, H. H.; McDonald, J. R. J. Chem. Phys. 1990, 93, 8777. (6) Hack, W.; Mill, Th. J. Phys. Chem. 1991, 95, 4712. (7) Hack, W.; Mill, Th. J. Phys. Chem. 1993,97, 5599. (8) Hack, W.; Rathmann, K. J. Phys. Chem. 1992,96,47. (9) Patel-Misra, D.; Dagdigian, P. J. J . Chem. Phys. 1992, 97, 4871. (10) Bohn, B.; Stuhl, F. J. Phys. Chem. 1993, 97,4891. (11) Hack, W.; Rathmann, K. J. Phys. Chem. 1992,96,47. (12) Adams, J. S.; Pasternack, L. J. Phys. Chem. 1991, 95, 2975. (13) Hawley, M.; Baronavski, A. P.; Nelson, H. H. J. Chem. Phys., to be published. (14) Rohrer, F. Ph.D. Dissertation,Ruhr-Universiat Bochum, Germany, 1987. (15) Freitag, F.; Rohrer, F.; Stuhl, F. J . Phys. Chem. 1989, 93, 3170. (16) Landolt-BBrnstein, part Sa, Transportphinomene, Springer, Berlin 1969. (17) Kenner, R. D.; Rohrer, F.;Stuhl, F. J. Phys. Chem. 1989,93,7824. (18) Piper, L. G.;Krech, R. H.; Taylor, R. L. J. Chem. Phys. 1980, 73, 791. Rohrer, F.; Stuhl, F. Chem. Phys. Let?. 1984, 111, 234. Hack, W.; Wilms, A. Z . Phys. Chem. 1989,161, 107. McDonald, J. R.; Miller, R. G.; Baronavski, A. P. Chem. Phys. Let?. 19i7,5i, 57. (22) Paur, R. J.; Bair, E. J. In?. J. Chem. Kiner. 1976,8, 139. (23) Cox, J. W.; Nelson, H. H.; McDonald, J. R. Chem. Phys. 1985,96, 175. ~. (24) Bower, R. D.; Jacoby, M. T.; Blauer, J. A. J. Chem. Phys. 1987,86, 1954. (25) Drozdoski, W. S.; Baronavski, A. P.; McDonald, J. R. Chem. Phys. Lett. 1979, 64, 421. (26) Nelson, H. H.; McDonald, J. R.; Alexander, M. H. J. Chem. Phys. 1990.94. 3291. (27) klexander, M. H.; Werner, H.-J.; Hemmer, T.; Knowles, P. J. J. Chem. Phys. 1990, 93, 3307. (28) Foy, B. R.; Casassa, M. P.; Stephenson, J. C.; King, D. S. J. Chem. Phys. 1990,92. 2782. (29) Casassa. M. P.; Foy, B. R.; Stephenson, J. C.; King, D. S.J . Chem. Phys. 1991, 94, 250. (30) Yarkony, D. R. J . Chem. Phys. 1990, 92,320. (31) Chen, J.; Quiilones, E.; Dagdigian, P. J. J. Chem. Phys. 1990, 93, 4033.