Professional Documents
Culture Documents
Spirko Bludsky Stabilizatiojn Van) Der Waals JMS 2000 PDF
Spirko Bludsky Stabilizatiojn Van) Der Waals JMS 2000 PDF
www.elsevier.nl/locate/molstruc
Abstract
The highly reliable potential energy function for the ground electronic state of the NaFH system based on the ab initio multi-
reference configuration interaction calculations [M.S. Topaler, D.G. Truhlar, X.Y. Chang, P. Piecuch, J.C. Polanyi, J. Chem.
Phys. 108 (1998) 5349; V. Špirko, P. Piecuch, O. Bludsky, J. Chem. Phys. 112 (2000) 189] is used to calculate the energies,
widths, and dissociation pathways of the quasi-bound states of the Na·· ·FH van der Waals molecule. Although emphasis is
placed on the quasi-bound states resulting from vibrational excitation of the HF fragment in the complex, the bound states and
the quasi-bound states, in which the HF fragment is not excited, are investigated as well. The study focuses on assessing the
reliability of the earlier stabilization calculations [V. Špirko, P. Piecuch, O. Bludsky, J. Chem. Phys. 112 (2000) 189] by
performing independent close-coupling calculations for the outgoing scattering waves of the NaFH system. The probabilities of
decay of the Na· · ·FH ro-vibrational resonances into various Na ⫹ HF
v; j channels provide new insights into the dynamics of
dissociation of the Na· ··FH complex. 䉷 2000 Elsevier Science B.V. All rights reserved.
Keywords: Quasi-bound states; van der Waals complexes; Close-coupling approach; Stabilization method; Ro-vibrational states
by 0.065 eV [5–7] or 0.076 eV [8]) is believed to tional problem with the stabilization method of
be the precursor of the photo-induced charge Mandelshtam et al. [25,26], we provided [5] the strong
transfer or “harpooning” [9,10] occurring in the evidence for the existence of long-lived ro-vibrational
electronically excited states of the complex, i.e. resonances corresponding to excitation of the H–F
[8,11,12], fragment in the complex to vr 1 and 2. According
to our stabilization calculations [5], the lifetimes of
Na…FH ⫹ hn !
Na…FHⴱ ‡ ! Na⫹ …
FH⫺ ‡ many vr 1 states of Na· · ·FH are in a nanosecond
regime. At least some vr 2 states have lifetimes in a
! NaF ⫹ H; (1) subnanosecond regime. By performing infinite-order
perturbative calculations based on the adiabatic
where the asterisk indicates an electronically separation of vibrational motions to define the unper-
excited complex and the double-dagger designates turbed problem [27,28], we demonstrated that the
an intermediate between reactants and products. long-lived resonances of Na· · ·FH are precisely those
Although recent quantum-scattering calculations that are bound in the adiabatic approximation. The
[13,14], based on accurate potential energy metastable states of Na· ··FH, which are not bound
surfaces (PESs) [8] that were obtained using the in the adiabatic approximation, are usually character-
multi-reference configuration interaction method ized by much shorter lifetimes.
with single and double excitations (the MRCISD There are two purposes of the present work. First of
approach; precisely, the MRDCI variant of the all, we would like to complete the earlier studies
MRCISD method [15–17]), indicate that the domi- [5,24] by providing information about dissociation
nant channel in the photo-induced Na…FH ⫹ hn pathways of the quasi-bound states of Na· ··FH.
process is photo-dissociation into reactants, Thus, we calculate the probabilities of decay of the
vr 0 and 1 states into the Na ⫹ HF
v; j channels,
Na…FH ⫹ hn !
Na…FHⴱ ‡ where v and j are, respectively, the vibrational and
rotational quantum numbers of the isolated HF
! Na ⫹ HF or Naⴱ ⫹ HF;
2 molecule. Calculations are performed using the
close-coupling approach [29,30] to the scattering
the same calculations indicate that there may be a states of the NaFH system. The second purpose of
possibility of improving the yield of products of this study is to verify the accuracy of our stabilization
reaction (1) by exciting the Na· · ·FH complex calculations of the vr 1 quasi-bound states of
vibrationally prior to electronic excitation. It is Na· · ·FH [5], which were based on the variational
well-known that the ground-state Na ⫹ HF ! NaF ⫹ treatment of the ro-vibrational eigenvalue problem,
H reaction also does not proceed without vibrational exci- by performing completely independent close-
tation of HF [18–23], which is a consequence of the coupling calculations of energy positions and widths
rather high energy barrier on the ground-state PES of the Na· · ·FH resonances. The stabilization method
of NaFH (1.26 eV [8]). used in Ref. [5] and the close-coupling approach to
Our recent calculations of the quasi-bound states of scattering states used in the present work are so
Na· ··FH [5], and a significant enhancement of the unrelated that the agreement of the results of both
infrared absorption line strengths corresponding to calculations can be used to argue that the information
the vr ⬎ 0 ← vr 0 transitions, when compared to obtained in these calculations is trustworthy and can
the isolated HF molecule [24] (vr is the quantum serve experimentalists in their studies of dynamics of
number for the H–F motion), suggest that one can Na· · ·FH. Indeed, in the stabilization method
contemplate an experiment in which the excitation employed in Ref. [5], the eigenvalues of energies of
of the H–F fragment in the complex prior to electronic resonance states are obtained by repeatedly calcu-
excitation is used to change the dynamics of the lating the eigenvalues of the ro-vibrational Hamilto-
Na…FH ⫹ hn process in favor of charge transfer, nian, while varying the size of the coordinate box in
Eq. (1). Indeed, by performing accurate calculations which the molecule is contained. A plot of the eigen-
combining the variational approach to the ro-vibra- values as a function of the size of the box (the
F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60 45
stabilization diagram) is used to extract the complex representation [8], to the following functional form,
resonance energies, X
V
r; R; u V 0 ⫹ C
a; b; g{exp⫺a1
r ⫺ rr }a
a ;b ;g
e Eres ⫺ iG=2;
3
{exp⫺a2
R ⫺ Rr }b Pg
cos u
where the real part (Eres) is the position of the reso-
nance and G is the width (i is the imaginary unit). The X
Ul
r; RPl
cos u; (5)
lifetimes of quasi-bound states, t , are then estimated l
through the relationship
where V0 ; C
a; b; g; a1, a2, rr, and Rr are the adjus-
t É=G;
4 table parameters and Pg (cos u ) designate the
Legendre polynomials. In order to improve the quality
where É h=
2p and h is the Planck constant. On the of PES, additional MRDCI calculations were
other hand, in the close-coupling approach, a system performed along the one-dimensional cuts of the adia-
of coupled one-dimensional Schrödinger equations in batic potential through the vdW minimum, corre-
the scattering coordinate is solved for the continuum sponding to fixing one of the three Jacobi
states of the Hamiltonian, which formally describe the coordinates at its equilibrium value, while allowing
scattering of Na from HF. The resulting collision-time the remaining two coordinates to vary over the entire
delay matrix provides information about the energy range. The details of these additional MRDCI calcu-
positions and widths of quasi-bound states and lations can be found in Ref. [5], whereas the details of
branching ratios for decay of these states into various the original MRDCI calculations used to construct a
Na ⫹ HF
v; j channels. 2 × 2 diabatic representation of the lowest two 2A 0
The paper is organized as follows. The ground-state states of NaFH are described in Ref. [8].
PES of the NaFH system is briefly discussed in Although the PES described in Ref. [5] could
Section 2. The stabilization and close-coupling basically be used in the present study, we decided to
approaches and the relevant computational details refit it. The reason for refitting the PES again is
are described in Section 3. The results of our calcula- primarily related to the fact that the close-coupling
tions are discussed in Section 4, and Section 5 calculations for scattering states require that the elec-
summarizes our findings. tronic potential V
r; R; u can be given the following
form,
^
V
r; R; u ⬅ V
r; R; r^ ·R
2. Potential energy surface
6
^
VH–F
r ⫹ VNa–FH
r; R; r^ ·R;
In describing the PES of Na· ··FH, which is used in
the stabilization and close-coupling calculations and ^ is the intermolecular potential
where VNa–FH
r; R; r^ ·R
in solving the corresponding ro-vibrational problem, obtained by subtracting the diatomic potential,
we use the internal Jacobi (or scattering) coordinates:
the H–F distance r ⬅ 兩r兩 (defining the high-frequency ^
VH–F
r lim V
r; R; r^ ·R;
7
R!∞
stretching coordinate), the distance R ⬅ 兩R兩 between
Na and the center of mass of HF (defining the vdW ^ (as usual, r^ and R
from V
r; R; r^ ·R ^ are the spherical
stretching or scattering coordinate), and the angle u angles of the Jacobi vectors, r and R, respectively).
between two Jacobi vectors, r and R (defining the This form of the potential, Eq. (6), is essential for the
bending or internal rotation coordinate). correct description of the boundary conditions for
The PES for the ground electronic state of the scattering states (coupled equations for partial scat-
NaFH system was described in our earlier paper [5]. tering waves used in the close-coupling calculations
It was obtained by refitting the adiabatic potential, should automatically decouple for R ! ∞; cf. Section
resulting from fitting highly accurate MRDCI ener- 3 for further details).
gies of the lowest two 2A 0 states to a 2 × 2 diabatic The PES reported in Ref. [5] can be given the form
46 F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60
Fig. 1. The PES for the ground electronic state of the NaFH system obtained by refitting the earlier MRDCI data [8] augmented with new
MRDCI energies [5] to Eq. (5), where C
a; b 0; g 0 for g 苷 0: The angle u used for plotting the PES is fixed at the value corresponding to
vdW minimum.
of Eq. (6), but only approximately [we can always try b , and g equal 6, 6, and 9, respectively, and 4 para-
to extract the diatomic potential VH–F
r from a given meters a1, a2, rr, and Rr]. These parameters are listed
potential V
r; R; u by studying the limit described by in Table 1. In constructing the new fit, we considered a
Eq. (7) for a fixed value of u assuming that, for suffi- wide range of almost 5000 nuclear geometries, with
ciently large R values, the potential V
r; R; u is energies up to 13,500 cm ⫺1 above the vdW minimum.
virtually isotropic]. In order to obtain the rigorous The differences between the fitted and calculated
decomposition of the total electronic potential MRDCI energies were usually 5–10 cm ⫺1 or smaller
^
V
r; R; u; Eq. (5), into VH–F
r and VNa–FH
r; R; r^ ·R; for energies up to ⬃1000 cm ⫺1 above the vdW
we must impose the following condition on the coeffi- minimum (this includes all bound and vr 0 quasi-
cients C
a; b; g; bound states). For energies up to 4500 cm ⫺1 above the
vdW minimum, including the important vr 1 quasi-
C
a; 0; g 0; for g 苷 0;
8
bound states, they usually were about 10–20 cm ⫺1.
so that V
r; R; u; Eq. (5), does not depend on u for The overall quality of the new fit is as good as in
R ! ∞: The above condition has not been imposed on the earlier work [5]. For example, the standard devia-
the potential reported in Ref. [5] and, in consequence, tion of the new fit is as small as in the case of the fit
V
r; R; u of Ref. [5] is only approximately indepen- described in Ref. [5] (ca. 60 cm ⫺1).
dent of u in the limit of infinite R. This had no effect All essential features of the ground-state PES of the
on the stabilization calculations performed in our NaFH system are accurately reproduced by the new
earlier study [5], which were based on enclosing the fit. In particular, the geometry of the vdW minimum,
NaFH system in the coordinate box along the vdW re 0:9207 A; Re 2:4720 A; ue 58:92⬚; and the
stretch coordinate R, but would cause problems in the binding energy of the Na· · ·FH complex, 0.079 eV
close-coupling calculations of quasi-bound states below the Na
3s 2 S ⫹ HF
X 1 S ⫹ asymptote, agree
reported here. very well with the original results reported in Ref. [8].
Thus, we refitted our earlier PES using the func- The slight disagreement with the binding energy
tional form described by Eq. (5), in which the coeffi- reported in Ref. [5] (0.065 eV) is a consequence of
cients C
a; b; g satisfy the condition (8). The refitted the fact that the earlier potential reported in Ref. [5]
PES, shown in Fig. 1, involves 125 parameters [121 does not satisfy the condition (8). The new fit predicts
coefficients C
a; b; g with the maximum values of a , the existence of the saddle point at almost the same
F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60 47
nuclear geometry and at almost the same energy as the values of Rmax is the well-known Sutcliffe–Tennyson
R 1:981 A;
original diabatic fit [8] (r 1:630 A; Hamiltonian defined in the body-fixed molecular
and u 81:6⬚; energy ⬃ 10; 200 cm⫺1 above the frame [34]. This Hamiltonian, designated here by
Na ⫹ HF asymptote). HSTJ
r; R; u; which parametrically depends on the
rotational quantum number J, after integrating over
the bending coordinate u and neglecting small
3. Dynamics of the Na ⫹ HF system below the Coriolis terms, reduces to an effective operator HST ⬅
0 0
reaction threshold: stabilization and close- 储具qk兩HSTJ
r; R; u兩q k 典储; which operates only on func-
coupling calculations of the vr ⱕ 1 quasi-bound tions of the radial variables r and R. We obtain,
states of the Na···FH molecule
具qk兩HST 兩q 0 k 0 典 ⫺É2 dq;q 0 dk;k 0
3.1. Stabilization calculations "
1 22 1 22
× ⫹ ⫺ q
q ⫹ 1
In the boxing variant [31,32] of the stabilization 2m2 2R2 2m1 2r 2
method [33], which we used in our earlier study [5] ( )
and use in this work (primarily to obtain new energy 1 1
⫹
positions resulting from refitting the earlier potential), 2m2 R2 2m 1 r 2
the energy positions and widths of resonance states #
are obtained by repeatedly calculating the eigenvalues J
J ⫹ 1 ⫺ 2k2
⫺
of the Hamiltonian, while gradually increasing the 2m 2 R 2
volume of the configuration space accessible to a X
molecule (defined by the coordinate box). As ⫹ dk;k 0 gl
q; q 0 ; kUl
r; R; (9)
mentioned in the Introduction, the complex energies l
Table 1
The parameters C
a; b; g defining the potential energy function V
r; R; u; Eq. (5), for the ground electronic state of NaFH (in cm ⫺1), used in
this paper. The remaining parameters characterizing V
r; R; u equal: a1 1:15 A ⫺1 ; a2 1:37 A
⫺1 ; rr 0:92 A
and Rr 2:47 A
equivalent to many more analytic functions, which are stretch coordinate R to get the converged description
often used in algebraic approaches to an eigenvalue of practically all vr ⱕ 1 quasi-bound states [5].
problem (e.g. eigenstates of harmonic oscillators), it is Without using the numerical basis functions for r
sufficient to use 4 numerical functions for the high- and R (particularly for the latter coordinate), we
frequency stretch coordinate r, 15 analytic functions would have significant problems in saturating the
[associated Legendre functions Q qk
u] for the bend, energy positions and widths of the vr 1 resonances
and as little as 60 numerical functions for the vdW with the basis set, since the basis set for R must be
F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60 49
i.e. the energy position Eres and the width G , are earlier [5], since we mainly focused on the long-
obtained by fitting the function Tr Q Jp
E [or lived vr 1 quasi-bound states, and for these states
Tr Qres;Jp
E] to the following formula: it is sufficient to use 4 numerical basis functions for
the H–F stretch, 15 basis functions for the bend
q
G ⫹ 2r
E ⫺ E res
Tr Qres;Jp
E É ⫹ C:
23 0–14; and 60 numerical basis functions for the vdW
E ⫺ Eres 2 ⫹ G 2 =4 stretch in order to obtain the converged values of Eres
The fitting procedure may start from determining the and G . The fact that we used the refitted potential,
maximum E0 in the resonance structure of Tr QJp
E described in Section 2, had practically no effect on
[or Tr Qres;Jp
E], followed by calculating the initial the calculated widths of the v r 1 resonances (within
estimate of the width through the formula G 0 the accuracy of the stabilization approach employed),
4É=Tr Qres;Jp
E0 : By calculating a few additional so that we could concentrate on the energy positions,
values of Tr Qres;Jp
E for the energies E around E0 which slightly changed compared to their earlier
(for example, at E E0 ^ G 0 =2; we can determine values [5] (typically, by 5–10 cm ⫺1; cf. Section 4).
all parameters of the fit (23), including Eres and G . The As in the earlier work [5], we varied the boxing para-
stabilization calculations described in Section 3.1, or meter Rmax between 6.5 and 9.0 Å with the steps 0.005
even simple estimates of the resonance energy or 0.01 Å.
positions obtained from perturbative calculations for As in the earlier paper [5], the assignment of
the quasi-bound states [5], are extremely useful in quantum numbers to states that are bound in the adia-
identifying the maxima E0 in the resonance structure batic approximation was achieved by performing a
of Tr Q Jp
E; particularly when the density of reso- two-step separation of the vibrational modes,
nances in a given energy range is large (as is the case followed by infinite-order perturbative calculations
for the vr 1 states of Na· ··FH). in which the zero-order problem is defined by the
The partial widths G i, for i 1; …; N open ; are deter- adiabatic approximation and the non-adiabatic
mined by fitting the individual diagonal elements of coupling terms between different vibrational motions
QJp
E [or Qres;Jp
E] to the formula, define the perturbation [27,28]. Let us recall that, in
the first step of this procedure, we separate the high-
Gi frequency H–F stretching mode from the low-
Qres;Jp
E É
i;i
E ⫺ Eres 2 ⫹ G 2 =4 frequency stretch and bend motions, obtaining a set
of smaller, two-dimensional effective problems corre-
⫹ Ci
0 ⫹ Ci
E ⫺ Eres :
24 sponding to different quantum states of the H–F
For the rotationless
J 0 quasi-bound states of stretch. In the second step, we replace each two-
Na· · ·FH, the index i in the above expression can be dimensional problem in R and u by a system of
identified with the ro-vibrational state of the HF coupled one-dimensional Schrödinger equations in R
molecule,
v; ji ; since, in this case, l j (cf. Ref. corresponding to different bending states, which can
[37]). Thus, the ratios G i =G; for i 1; …; N open ; give be solved variationally or using the close-coupling
us immediately the branching ratios for decay of the approach [5]. The quantum numbers for the high-
J 0 quasi-bound states of the Na· · ·FH complex into frequency stretch, the bend, and the vdW stretch,
various Na ⫹ HF
v; j channels. resulting from the above procedure, are designated
Although in the above description, we focused on by vr, vu , and vR, respectively.
the close-coupling calculations for the quasi-bound The ro-vibrational states of HF,
v; j 0–jmax ;
used to define the basis set functions F kJMp
r; R;^
states, information about the bound-states can be
obtained from these calculations as well. We refer Eq. (16), and to construct the corresponding close-
the reader to Ref. [37] for further details. coupling expansion (17), were the following: (0,0–
19), (1,0–19), (2,0–17), and (3,0–15). The energies
3.3. The remaining computational details e
v; j and the corresponding functions x
v; j
r were
determined from the HF potential VH–F
r; Eq. (7)
The stabilization calculations performed in this (the potential of HF extracted from the PES of
study were somewhat easier than those reported NaFH). The system of 74 coupled equations, Eq.
52 F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60
Table 5
The decay probabilities for the vr 1 quasi-bound vibrational
J 0 states of the Na···FH vdW molecule into the Na ⫹ HF
v 0; j channels
resulting from the close-coupling calculations using 74 basis functions F kJMp
r; R; ^ Eq. (16). The decay probabilities were obtained by
calculating the ratios G i =G for all open channels i (see the text for further details). The Na…FH
v r 1; vu ; vR states below the Na ⫹ HF
v
1; j 0 threshold (predominantly, the Feshbach-type resonances). The meaning of vr, vu , and vR is the same as in Table 2
(vr, vu , vR)
v j (1,0,0) (1,0,1) (1,0,2) (1,0,3) (1,0,4) (1,0,5) (1,0,6) (1,1,0) (1,1,1) (1,2,0) (1,1,2) (1,1,3)
0 0 0.001 0.002 0.003 0.006 0.005 0.005 0.012 0.000 0.001 0.001 0.002 0.001
1 0.003 0.004 0.007 0.016 0.012 0.012 0.031 0.000 0.002 0.004 0.004 0.003
2 0.003 0.004 0.007 0.015 0.012 0.012 0.032 0.000 0.002 0.003 0.005 0.003
3 0.004 0.005 0.009 0.023 0.017 0.017 0.049 0.000 0.002 0.004 0.009 0.004
4 0.011 0.013 0.023 0.058 0.041 0.042 0.119 0.001 0.003 0.011 0.023 0.009
5 0.019 0.020 0.037 0.098 0.066 0.067 0.198 0.000 0.004 0.016 0.045 0.014
6 0.016 0.016 0.032 0.104 0.061 0.063 0.242 0.000 0.001 0.012 0.087 0.011
7 0.000 0.001 0.003 0.025 0.007 0.009 0.145 0.012 0.019 0.010 0.163 0.023
8 0.119 0.139 0.166 0.096 0.158 0.142 0.013 0.076 0.190 0.166 0.241 0.220
9 0.309 0.308 0.405 0.382 0.455 0.371 0.112 0.086 0.274 0.271 0.115 0.341
10 0.492 0.466 0.298 0.165 0.110 0.251 0.033 0.717 0.460 0.480 0.253 0.350
11 0.021 0.025 0.012 0.010 0.055 0.010 0.007 0.111 0.042 0.021 0.048 0.021
12 0.000 0.000 0.000 0.001 0.001 0.001 0.003 0.000 0.001 0.000 0.005 0.001
13 – – 0.000 0.000 0.001 0.000 0.002 0.000 0.000 0.000 0.000 0.001
the stabilization and close-coupling calculations for states listed at the bottom of Table 4 and obtained in
the vr 1 quasi-bound states, which are bound in the close-coupling calculations [the
1; x; x 0 ;
1; y; y 0 ;
the adiabatic approximation (both calculations using and
1; z; z 0 states]. We could not assign quantum
the same PES), are virtually identical (see Table 4). numbers vr, vu , and vR to these states, so that they
The energy positions resulting from both calculations are, most likely, not bound in the adiabatic approxi-
differ by 1 cm ⫺1 or less. The agreement between the mation. As in the case of some vr 0 resonances, we
widths obtained in both calculations is excellent as could not precisely determine which higher-energy
well. The fact that two entirely different mathematical vr 1 quasi-bound states identified in the earlier
formalisms, the stabilization approach, which is stabilization calculations [5] correspond to states
strongly tied to calculations of the ro-vibrational
1; x; x 0 ;
1; y; y 0 ; and
1; z; z 0 obtained in the present
bound states, and the close-coupling approach, in close-coupling calculations (this is why we left blank
which we directly solve the Schrödinger equation spaces in Table 4). At this point, we can only spec-
for the continuum states, yield almost identical reso- ulate that the vr 1 state with the energy Eres
nance energies and widths gives credence to our 4550:48 cm⫺1 above the Na ⫹ HF
v 0; j 0
earlier prediction of the existence of the long-lived dissociation threshold [designated in Table 4 as
quasi-bound states of Na· ··FH corresponding to exci-
1; z; z 0 ; located at 4994.15 cm ⫺1 above the lowest
tation of the HF fragment to vr 1. bound state; cf. Tables 2 and 4] corresponds to the
It seems that the very good agreement between the vr 1 resonance, stabilized in the previous study
results of stabilization and close-coupling calculations [5], with the energy Eres 4995:7 cm⫺1 above the
is obtained only for the long-lived vr 1 resonances, (0,0,0) state (see Table 4 in Ref. [5]). The latter
which are characterized by rather small widths G and state has the width G equal to 0.345 cm ⫺1, and the
which are bound in the adiabatic approximation. state with E res 4550:48 cm⫺1 obtained in the
Larger differences between the calculated resonance present close-coupling calculations has G equal to
energies and widths are observed for the higher- 0.214 cm ⫺1 (see Table 4). Although the remaining
energy v r 1 states, which are short-lived and two states at the bottom of Table 4 [the
1; x; x 0 and
which are, most likely, not bound in the adiabatic
1; y; y 0 states] have energies that look very similar to
approximation. Examples of such states are the three energies of some high-energy v r 1 states reported
56 F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60
Table 6
Same as Table 4 for the Na…FH
vr 1; vu ; vR quasi-bound states
with J 0 above the Na ⫹ HF(v 1, j 0) threshold
(vr, vu , vR)
The fact that the rotationless vr 1 states, which izing the product HF
v 0; j molecules and the
are bound in the adiabatic approximation, lie below quantum numbers characterizing the vr 1 quasi-
the Na ⫹ HF
v 1; j 0 threshold and that the J bound states, which are bound in the adiabatic approx-
0; vr 1 states, which are not bound in the adiabatic imation (see Table 5 and Fig. 2). It seems that excita-
approximation, have energies above the Na ⫹ HF
v tion of the vdW stretching mode (from vR 0 to 6)
1; j 0 threshold has an effect on the dissociation broadens the range of j values characterizing the final
pathways of the vr 1 states. The vr 1 states, Na ⫹ HF
v 0; j channels and shifts the average j
which are bound in the adiabatic approximation, value towards j 0; from three values of j
j
decay into the Na ⫹ HF
v 0; j channels (see 8–10 that give the significant decay probabilities
Table 5). The remaining vr 1 states decay into the for vR 0 to about nine values of j
j 1–7; 9; 10
Na ⫹ HF
v 1; j and Na ⫹ HF
v 0; j channels that essentially contribute to the rotational product
(see Table 6). distribution for vR 6 [cf. Table 5 and Fig. 2a]. Even-
These different dissociation pathways of the vr 1 tually, when vR ⬎ 6; so that the corresponding vr 1
states below and above the Na ⫹ HF
v 1; j 0 states of Na· ··FH are no longer bound in the adiabatic
threshold have interesting implications on the approximation, we observe a significant redistribution
observed distributions of the final j values. Indeed, of the final j values, with j 0 or 1 being now the
since the vr 1 states of Na· · ·FH, which are bound dominant channels, since most of the energy released
in the adiabatic approximation, cannot decay into the in the decay of the Na…FH
vr 1; J 0 states,
Na ⫹ HF
v 1; j channels, the large amount of the which lie above the Na ⫹ HF
v 1; j 0 threshold,
vibrational energy stored in these states (ca. remains stored in the vibrationally excited, HF
v
4000 cm ⫺1) must partially be converted into signifi- 1; j; molecules. It seems that excitation of the bend
cant rotational excitation of the HF product molecule. has a small effect on the final rotational product distri-
The results in Table 5 confirm this expectation (see, bution. Larger effect is observed only when excitation
also, Fig. 2). The rotationally excited HF
v 0; j of the bend is accompanied by excitation of the vdW
molecules, with j as high as 9–11, are the dominant stretch [as in the case of the (1,1,2) state, which
products of the decay of the Na…FH
vr 1; J 0 dissociates into Na ⫹ HF
v 0; j channels with j
resonances below the Na ⫹ HF
v 1; j 0 4–11; cf. Table 5 and Fig. 2b].
threshold. On the other hand, the lowest energy
Na…FH
vr 1; J 0 states, which are not bound
in the adiabatic approximation and lie above the Na ⫹ 5. Summary and concluding remarks
HF
v 1; j 0 threshold [the
1; x; x 0 ;
1; y; y 0 ;
and
1; z; z 0 states in Tables 4 and 6] decay primarily A highly reliable analytical representation of the
into the Na ⫹ HF
v 1; j states with low j values potential energy function for the ground electronic
[e.g. j 0 in the case of the
1; x; x 0 state, or j state of the NaFH system was used to evaluate the
0; 1 in the case of the
1; y; y 0 state]. The HF
v energies and widths of the quasi-bound ro-vibrational
1; j molecules with larger j values [e.g. j 3 in the states of the Na· · ·FH vdW complex. An emphasis was
case of the
1; z; z 0 state] are produced only when the placed on a comparison of the previously published
energy Eres of a given vr 1 resonance is significantly [5] and newly obtained results of the stabilization
above the Na ⫹ HF
v 1; j 0 threshold. This calculations with the results of the converged close-
resembles the decay patterns observed for the vr 0 coupling calculations. The probabilities of decay of
resonances, which decay into the Na ⫹ HF
v 0; j the quasi-bound states of Na· · ·FH into individual
channels with small j values only if their energy posi- Na ⫹ HF
v; j channels were obtained using the
tions are just above the corresponding Na ⫹ HF
v close-coupling approach. Although we focused on
0; j 0 dissociation threshold [for other vr 0 reso- the quasi-bound states, in which the HF fragment in
nances, with larger energies Eres, the rotationally the complex is excited to vr 1; information about
excited HF
v 0; j molecules are obtained]. the lower-energy vr 0 resonances and bound states
Finally, we would like to comment on the relation- was obtained and analyzed as well.
ship between the distribution of j values character- It was demonstrated that the converged stabilization
58 F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60
calculations, using the approach developed by values characterizing the Na ⫹ HF
v 0; j channels,
Mandelshtam et al. [25,26] and exploited earlier in into which the Na…FH
vr 1; J 0) states that are
Ref. [5], and the closed-coupling calculations, based bound in the adiabatic approximation decay.
on the ideas and algorithms described in Refs. In the future, we hope to be able to extend our
[30,37,41,42] and in Section 3 of this paper, lead to close-coupling calculations to vr 2 states, at least
virtually identical results for all bound and vr 1 to those vr 2 states, which are bound in the adia-
quasi-bound states, which are bound in the adiabatic batic approximation [below the Na
3s 2 S ⫹
approximation. The results of the close-coupling HF
X 1 S ⫹ ; v 2; j 0 threshold]. Information
calculations for the vr 0 resonances, which do not about some of these states has already been provided
lie in the immediate vicinity of the Na(3s 2S) ⫹ in our earlier work [5], and we hope that this informa-
HF(X 1S ⫹; v 0, j 0) dissociation threshold, also tion will facilitate the close-coupling calculations for
turned out to be very similar to those obtained with the the vr 2 states. The close-coupling calculations
stabilization method. Larger differences between both become a lot easier, if the approximate energy posi-
types of calculations were observed only for the tions of the quasi-bound states of interest are a priori
v r 1 resonances above the Na
3s 2 S ⫹ HF
X 1 S ⫹ ; known.
v 1; j 0 threshold and for the vr 0 resonances The detailed information about energies, widths,
in the immediate vicinity of the Na
3s 2 S ⫹ and dissociation pathways of the vr 1 and 2 reso-
HF
X 1 S ⫹ ; v 0; j 0 threshold. nances provided here and in Ref. [5], particularly
The excellent agreement between the fully about the states, which are characterized by the longer
converged close-coupling and stabilization calcula- lifetimes, may play a critical role in assessing the
tions for the vr 1 quasi-bound states, which are feasibility of the experiment mentioned in Section 1
bound in the adiabatic approximation, means that and in Refs. [5,24] in which excitation of the H–F
we can rely on either method, if we are to provide stretching mode in the complex prior to electronic
the most essential characteristics (energy positions excitation of Na· · ·FH could be used to explore new
and widths) of these states. In particular, our earlier dynamical scenarios for the photo-induced charge
claim that there exist many long-lived resonances that transfer in vdW molecules. Information about the
correspond to excitation of the H–F mode in the vr ⬎ 0 quasi-bound states may also be useful in accu-
complex to vr 1 has now a very solid support in rate state-to-state description of the ground-state
the form of the results of very accurate close-coupling Na ⫹ HF ! NaF ⫹ H reaction [22,23] (for the
calculations. experimental studies of this reaction, see Refs. [18–
We noticed that practically all vr 1 states of 20]). The vr 0 resonances are essential for the
Na· ··FH, which are bound in the adiabatic approxima- understanding of the non-reactive scattering of the
tion, have energies below the Na
3s 2 S ⫹ Na atoms from the HF
v 0; j molecules [6,7].
HF
X 1 S ⫹ ; v 1; j 0 dissociation threshold. The fact that we now have a reliable theoretical
The vr 1 states above the Na
3s 2 S ⫹ methodology at hand that allows us to extract reason-
HF
X 1 S ⫹ ; v 1; j 0 threshold are not bound in ably accurate information about the ro-vibrational
the adiabatic approximation. Clearly, a similar remark resonances of the Na· ··FH vdW molecule means
applies to the vr 0 resonances, which must necessa- that, in the future, we may be able to provide a great
rily lie above the corresponding Na
3s 2 S ⫹ deal of information about the dynamics of other
HF
X 1 S ⫹ ; v 0; j 0 threshold (otherwise, the complexes formed by the alkali and alkaline-earth
vr 0 states would be the bound states). metal atoms and halides, which have recently been
We demonstrated that the vr 1 quasi-bound studied in several laboratories [11,43–59].
states, which are bound in the adiabatic approxima-
tion, decay into the Na ⫹ HF
v 0; j channels with
large j values. The states above the Na
3s 2 S ⫹ Acknowledgments
HF
X 1 S ⫹ ; v 1; j 0 threshold primarily decay
into the Na ⫹ HF
v 1; j channels. The excitation One of us (PP) would like to thank Profs Henryk
of the vdW stretching mode increases the range of j Ratajczak and Austin Barnes for the invitation to
F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60 59
contribute to this special issue of the Journal of [13] Y. Zeiri, G. Katz, R. Kosloff, M.S. Topaler, D.G. Truhlar,
Molecular Structure. He would also like to thank J.C. Polanyi, Chem. Phys. Lett. 300 (1999) 523.
[14] G. Katz, R. Kosloff, A. Jasper, M.D. Hack, M.S. Topaler,
Profs Henryk Ratajczak and Jan Baran for inviting
D.G. Truhlar, P. Piecuch, J.C. Polanyi, unpublished.
him to speak at the “Vth International Conference [15] R.J. Buenker, S.D. Peyerimhoff, Theor. Chim. Acta 35 (1974)
on Molecular Spectroscopy.” This work has been 33.
supported by the startup funds provided to one of us [16] R.J. Buenker, S.D. Peyerimhoff, Theor. Chim. Acta 39 (1975)
(PP) by the Michigan State University (MSU) and by 217.
[17] P.J. Bruna, S.D. Peyerimhoff, Adv. Chem. Phys. 67 (1998) 1.
the MSU Intramural Research Grant Program (the
[18] B.A. Blackwell, J.C. Polanyi, J.J. Sloan, (communicated
New Faculty, Science and Engineering Award during the General Discussion at the Faraday Discussion of
received by PP). Some of the calculations reported the Chemical Society), Faraday Discuss. Chem. Soc. 62
in this paper (parallel calculations of stabilization (1977) 147.
diagrams) were performed on the R12000 based 128 [19] B.A. Blackwell, J.C. Polanyi, J.J. Sloan, Chem. Phys. 30
processor Origin 2000 supercomputer in Eagan, (1978) 299.
[20] F.E. Bartoszek, B.A. Blackwell, J.C. Polanyi, J.J. Sloan, J.
Minnesota. We would like to thank Silicon Graphics Chem. Phys. 74 (1981) 3400.
Incorporated for providing one of us (PP) with an [21] A. Laganá, M.L. Hernandez, J.M. Alvariño, L. Castro,
access to this resource. Additional support by the P. Palmieri, Chem. Phys. Lett. 202 (1993) 284.
Grant Agency of the Academy of Sciences of the [22] A. Laganà, J.M. Alvariño, M.L. Hernandez, P. Palmieri,
Czech Republic (Grant No. A4040806) is gratefully E. Garcia, T. Martinez, J. Chem. Phys. 106 (1997) 10 222.
[23] R. Gargano, S. Crocchianti, A. Laganà, G.A. Parker, J. Chem.
acknowledged. Finally, one of us (PP) would like to
Phys. 108 (1998) 6266.
thank Prof. John C. Polanyi and Dr Andrew J. Hudson [24] M.S. Topaler, P. Piecuch, D.G. Truhlar, J. Chem. Phys. 110
for many stimulating discussions and considerable (1999) 5634.
interest in the results reported in this paper. [25] V.A. Mandelshtam, T.R. Ravuri, H.S. Taylor, Phys. Rev. Lett.
70 (1993) 1932.
[26] V.A. Mandelshtam, H.S. Taylor, V. Ryaboy, N. Moiseyev,
Phys. Rev. A 50 (1994) 2764.
References [27] V. Špirko, J. Cı́žek, J. Chem. Phys. 102 (1995) 8906.
[28] V.Špirko, M. Rozložnı́k, J. Cı́žek, Phys. Rev. A 61 (2000)
[1] K.R. Leopold, G.T. Fraser, S.E. Novick, W. Klemperer, 014102.
Chem. Rev. 94 (1994) 1807. [29] A.C. Allison, Adv. At. Mol. Phys. 25 (1988) 323.
[2] A. van der Avoird, P.E.S. Wormer, R. Moszyński, Chem. Rev. [30] F. Mrugała, Int. Rev. Phys. Chem. 12 (1993) 1.
94 (1994) 1931. [31] R. Lefebvre, J. Phys. Chem. 89 (1985) 4201.
[3] R.E. Miller, in: E.R. Bernstein (Ed.), Chemical Reactions in [32] M. Garcia-Sucre, R. Lefebvre, Chem. Phys. Lett. 130 (1986)
Clusters, Oxford University Press, New York, 1996, pp. 40– 240.
63. [33] A.U. Hazi, H.S. Taylor, Phys. Rev. A 1 (1970) 1109.
[4] J.A. Beswick, in: A. Weber (Ed.), Structure and Dynamics of [34] B.T. Sutcliffe, J. Tennyson, Mol. Phys. 58 (1986) 1053.
Weakly Bound Molecular Complexes, Reidel, Dordrecht, [35] E.U. Condon, G.H. Shortley, The Theory of Atomic Spectra,
1987, pp. 563–571. Cambridge University Press, Cambridge, 1963.
[5] V. Špirko, P. Piecuch, O. Bludsky, J. Chem. Phys. 112 (2000) [36] J.W. Cooley, Math. Comput. 15 (1961) 363.
189. [37] F. Mrugała, R. Moszyński, J. Chem. Phys. 109 (1998) 10 823.
[6] R. Düren, U. Lackschewitz, S. Milošević, H.J. Waldapfel, J. [38] W.A. Lester Jr., in: W.H. Miller (Ed.), Modern Theoretical
Chem. Soc., Faraday Trans. II 85 (1989) 1017. Chemistry, vol. 1, Plenum, New York, 1976, p. 1.
[7] R. Düren, (communicated during the General Discussion at [39] W.J. Thompson, Angular Momentum, Wiley, New York,
the Faraday Discussion of the Chemical Society), J. Chem. 1994.
Soc., Faraday Trans. II 85 (1989) 1050. [40] F.T. Smith, Phys. Rev. 118 (1960) 349.
[8] M.S. Topaler, D.G. Truhlar, X.Y. Chang, P. Piecuch, [41] F. Mrugała, J. Chem. Phys. 91 (1989) 874.
J.C. Polanyi, J. Chem. Phys. 108 (1998) 5349. [42] F. Mrugala, J. Chem. Phys. 93 (1989) 1257.
[9] M. Polanyi, Atomic Reactions, Williams and Norgate, [43] B. Soep, C.J. Whitham, A. Keller, J.P. Visticot, Faraday
London, 1932. Discuss. Chem. Soc. 91 (1991) 191.
[10] D.R. Herschbach, Adv. Chem. Phys. 10 (1966) 319. [44] B. Soep, S. Abbés, A. Keller, J.P. Visticot, J. Chem. Phys. 96
[11] X.Y. Chang, R. Ehlich, A.J. Hudson, P. Piecuch, J.C. Polanyi, (1991) 440.
Faraday Discuss. Chem. Soc. 108 (1997) 411. [45] A. Keller, R. Lawruszczuk, B. Soep, J.P. Visticot, J. Chem.
[12] M.S. Topaler, D.G. Truhlar, X.Y. Chang, P. Piecuch, Phys. 105 (1996) 4556.
J.C. Polanyi, J. Chem. Phys. 108 (1998) 5378. [46] B. Soep, (communicated during the General Discussion at the
60 F. Mrugała et al. / Journal of Molecular Structure 555 (2000) 43–60
Faraday Discussion of the Chemical Society), Faraday [54] S. Skowronek, R. Pereira, A. Gonzalez-Ureña, J. Phys. Chem.
Discuss. Chem. Soc. 108 (1997) 459. A 101 (1997) 7468.
[47] R. Lawruszczuk, M. Elhanine, B. Soep, J. Chem. Phys. 108 [55] V. Stert, P. Farmanara, W. Radloff, F. Noack, S. Skowronek,
(1998) 8374. J. Jimenez, A. Gonzalez-Ureña, Phys. Rev. A 59 (1999)
[48] K. Liu, J.C. Polanyi, S.H. Yang, J. Chem. Phys. 96 (1992) R1727.
8628. [56] S. Skowronek, J.B. Jimenez, A. Gonzalez-Ureña, Chem. Phys.
[49] K. Liu, J.C. Polanyi, S.H. Yang, J. Chem. Phys. 98 (1993) Lett. 303 (1999) 275.
5431. [57] P. Farmanara, V. Stert, W. Radloff, S. Skowronek, A.
[50] J.C. Polanyi, J.-X. Wang, S.H. Yang, Isr. J. Chem. 34 (1994) Gonzalez-Ureña, Chem. Phys. Lett. 304 (1999) 127.
55. [58] S. Skowronek, J.B. Jimenez, A. Gonzalez-Ureña, J. Chem.
[51] J.C. Polanyi, J.-X. Wang, J. Phys. Chem. 99 (1995) 13 691. Phys. 111 (1999) 460.
[52] X.Y. Chang, R. Ehlich, A.J. Hudson, J.C. Polanyi, J.-X. Wang, [59] S. Skowronek, A. Gonzalez-Ureña, Prog. React. Kinet. Mech.
J. Chem. Phys. 106 (1997) 3988. 24 (1999) 101.
[53] S. Skowronek, R. Pereira, A. Gonzalez-Ureña, J. Chem. Phys.
107 (1997) 1668.