[Legenda] 

[Contents of the thesis] 

[Comments on this module] 



[Legenda] 

[Contents of the thesis] 

[Comments on this module] 
By comparison of measurements and calculations the missing parameters of the covalent potential curve and the value of H_{12} can be estimated. Moreover, the suitability of the applied stationaryphase approximation and LandauZener theory can be rated at its true value because the ionic potential curve is well known.
Fig. A08m5bii1F1
a shows the polar differential cross section, defined by
, for the chemiionization process of sodium on iodine and calculated by the lowestorder stationaryphase approximation .
Use has
been made of the LandauZener formula Eqs. (E1)


Figure: A08m5bii1F1: Polar differential cross section for chemiionization (CM
system) at E_{i}=13.1 eV,
calculated in semiclassical approximation with the potential
parameters of Table A08m5biT1 and the coupling parameters H_{12}=
0.065 eV and H_{rot}=0 eV. (a) Differential cross section with complete interference structure, calculated with the lowestorder stationaryphase approximation. The region of the classical rainbow angle _{cl,rb} has been omitted. (b) Differential cross section calculated with the stationaryphase approximation and uniform rainbow approximation showing separated the longwavelength interference structures due to a + c ( 0 65, full curve), b + c ( 65 250, full curve) and d + e ( 0 2300, dashed curve) interferences. (c) Full bars indicate the measured maxima of the interference structure on the differential cross section due to netattractive scattering. Dashed bars indicate the maxima due to netrepulsive scattering. The complete measured cross section curve at E_{r} =13.1 eV is given in Fig. 
Fig. A08m5bii1F1b shows the differential cross section with simplified interference structure. The major part has been calculated by the lowestorder stationaryphase approximation
The lowestorder stationaryphase approximation is used in the form of Eq. (E5)
(where
indicates the phase shift and k is given by ) and Eq. (6).


The rainbow has been calculated using Eq. (E7).
 
Because the origin of the oscillatory features of Fig. A08m5bii1F1a can be seen easily from Fig. A08m5bii1F1b and the latter figure is more easily related to the deflection function and the potential curves, Fig. A08m5bii1F1b is more suitable to fit the interference structure of the calculated and measured differential cross section by adjustment of the potential parameters. Moreover, convolution of the differential cross section of Fig. A08m5bii1F1a and Fig. A08m5bii1F1b even with the smallest angular resolution of the detector of 0.3º fwhm will lead to equal results.
The rainbow structure of the experimental cross sections is explained nearly perfectly by the theoretical description.
Figure A08m5bii1F3: Rainbow structure versus initial colliding energy (CM system). The dashed curves show the calculated positions of the maxima of rainbow and supernumeraries 
The ionic potential curve Na^{+}Cl^{} is well known. Moreover, the energy difference
of ionic and covalent ground state at infinite separation is known and the
covalent potential must be about zero for
.
So the ionic
scattering takes place via wellknown potential curves leading to a
complete knowledge of the relevant deflection curve (branches b, c
and e of Fig. A08m5biF1).
By applying the stationaryphase approximation and the uniform rainbow approximation, the positions of primary and supernumerary rainbows from interference of the b and c contributions are calculated. Fig. A08m5bii1F3 compares the measured and calculated relative positions of the rainbow maxima, where the measured as well as the calculated third supernumeraries have been placed on the same straight line. ReliabilityFrom the primary rainbows up to the third supernumeraries the agreement is nearly perfect. Further up an increasing deviation is observed of measured and calculated spacings of the successive supernumeraries. Moreover, the deviation increases with increasing collision energy. For 13.1 eV the deviation increases from 5 up to 20 percent while for the rainbow structure at 38.7 eV the corresponding quantities are 5 and 40 percent. A look at the potential and deflection curves shows that even for the 13.1 eV deviation, a reasonable adjustment of the potential parameters is not able to decrease the distance between the b and c branches of the deflection curve enough to fit the measured and calculated rainbow structures. Moreover, an energydependent deviation cannot be eliminated in this way. 

The cross sections exhibit Stueckelberg oscillations due to interference of the ionic and covalent repulsivescattering contributions just beyond the rainbow angle. The corresponding branches of the deflection curve are d and e, where the ebranch is calculated from the known ionic potential.
The Stueckelberg oscillations in the experimental cross sections shown in Fig. A08m5bii1_F2 for < 65 eV degree are due to the interference of the differential crosssection contributions related to the a and c branches of the deflection curves. The measured maxima of the 13.1 eV curve have been indicated in Fig. A08m5bii1F1 c together with the calculated structure in Fig. b.
For Stueckelberg oscillations just beyond the rainbow angle, the measured wavelength as a function of the kinetic energy can be fitted perfectly with the calculations.
However, on the range 65 obviously the distances between the calculated maxima are too small. The relative wavelength discrepancy is the largest at .
For decreasing scattering angles the measured and calculated oscillation wavelengths increase while the deviation decreases from twenty percent at to zero percent at and ends up in an opposite deviation meaning a calculated wavelength too large compared to the measurements. However, the latter deviation is not precarious because it can be improved by a careful adjustment of the onset of the covalent repulsive potential as can be seen from Fig. A08m5biF4 of the potential and Fig. A08m5biF4 of the deflection function.
More serious is the deviation on the 65 range because this part of the differential cross section deals only with scattering from the known ionic potential and the flat part of the covalent potential curve. It is very remarkable that the deviation on this range is the same as the deviation of measured and calculated maxima just on the other side of the minimum in the differential cross section (the range of the highnumber supernumeraries). The completely resolved differential cross section curve with E_{i}=18.2 eV gives the same trend with equal deviations of measured and calculated wavelengths around = 65 eV degree.
Some evidence of repulsive interference at small scattering angles has been indicated also in Fig. A08m5bii1F1c. This structure gives a check on the estimated repulsive part of the covalent potential. Because Fig. A08m5bii1F1 b clearly shows that at small angles the repulsive oscillation is the straight continuation of the attractive oscillation it is not very surprising that the calculated wavelength is somewhat too large compared to the measurements. The same applies to the 18.2 eV curve.
The general shape of the polar differential cross section is the cross section averaged over the quantal oscillations. This shape will nearly correspond to the classical differential cross section except at the classical rainbow angle, and thus is given by the relevant values of b, and the LandauZener transition probability P_{b} .
The general shape of the differential cross section has been measured and calculated at collision energies of 13.1, 20.7, 29.7, 38.7 and 55.0 eV, as shown in Figs. A08m5bii1F4a, b, c, d, e. The calculated values have been given on absolute scales, the measurements are only relative and have been given for the different energies on arbitrary, nonrelated scales.
The angular positions of special features agree very well: namely the maximum of the peak at due to covalent scattering, the minimum at due to scattering with maximal impact parameter and the maximum of the primary rainbow.
A permanent local disagreement is observed at the minimum of the differential cross section at , showing that the calculated cross section is too small continuously. The deviation increases at increasing energy. At least the major part cannot be caused by convolution effects on the measurements.
It has been shown that scattering from wellknown parts of the internuclear potentials gives rise sometimes to a discrepancy of the calculated and measured interference structure. In spite of that, we have determined the repulsive parameters of the covalent potential curve from the repulsive interference structure.
However, the reliability is enlarged by the very good agreement of the kineticenergy behaviour of the wavelength , while the disagreement of the interference wavelength on the range 0 150 shows an energydependent discrepancy. Perhaps the collisions with large impact parameters cause this discrepancy. The comparable calculated and measured interference structures due to collisions with smaller impact parameters consist of the primary rainbows and lownumber supernumeraries that are in very good agreement.
In the preceding we have seen that there are only a few serious deviations of measurements and calculations that cannot be ascribed to measuring faults or doubts as to the correctness of the potentials used. Summarizing, these discrepancies are:
The deviations must be viewed in the light of the used approximations, i.e. in the light of the restrictions of the theoretical methods we used ,the most important of which are:
Comparing figures A08m4bi1F1 and A08m5bii1F2, we see that the deviations between measurements and calculations occur only near = 65 eV degree, for covalent as well as ionic scattering. The deviations are energy dependent and deal with the oscillation wavelengths and relative intensities. It is very remarkable that the approximations C, D, E, F and G mostly violate the real differential cross sections at the region around = 65.
Those approximations don't change the oscillatory structure but only the relative intensities. Some estimates of corrections to the approximations have been made, especially in view of the influence on the differential cross section at .
A correction to approximation D (The use of the LandauZener formula to collisions where the distance of closest approach R_{0} and the distance of pseudocrossing R_{c} are not well separated.) reduces P_{b} by passing an incomplete transition region. Because then in our energy range the product P_{b}(1  P_{b}) decreases too, the differential cross section even decreases in the region of interest.
However, a correction to E (The use of the diabatic potentials in the classical deflection function in spite of small deformation of the curves at the pseudocrossing.) indeed predicts a somewhat larger value of the differential cross section close to = 65, but there still exists the feature of for .
It is very difficult to introduce a correction to approximation F (The use of a transition point in spite of a transition region around the pseudocrossing predicted by the LandauZener theory]. in the impactparameter method. It will lead to a collection of deflection curves with the greatest differences of the relative shapes near . Indeed, a summation over the deflection curves never gives a zero value of the differential cross section due to collisions with large impact parameters. However, this averaging effect does not cause an important rise of the cross section on the whole region around = 65.
The remaining corrections to the intensity at are C and G dealing with the LandauZener coupling and rotation coupling. It is not expected that the LandauZener transition formula gives such wrong results, although this formula has been derived using the approximations that U_{ion} and U_{cov} are linear in the region of the crossing and H_{12} is an essentially constant coupling element.
We have greater expectations for the rotation coupling to explain the intensity deviation. Rotational coupling will be treated in more detail in module A08m5bii2 Quantitative interpretation .
If the rotation coupling is taken into account as well, the only one of the approximations AG that can give a wrong result for the oscillatory wavelength of the differential cross section at is the lowestorder stationaryphase approximation. Delos and Thorson have given four statements that justify the application of this approximation on twostate collisions. One of them should not be satisfied in our case, namely the requirement of a collisional energy large compared to the potentialenergy differences of the two states. This statement requires about equal classical trajectories along the two states. The deflection curves of Fig. A08m5biF2 show that this requirement is not fulfilled.