Theoretical study of the absorption spectra of the lithium dimer



                            H.-K. Chung, K. Kirby, and J. F. Babb

                     Institute for Theoretical Atomic and Molecular Physics,

                         Harvard-Smithsonian Center for Astrophysics,

                            60 Garden Street, Cambridge, MA 02138



                                            Abstract



           For the lithium dimer we calculate cross sections for absorption of radi-

          ation from the vibrational-rotational levels of the ground X 1+g electronic

          state to the vibrational levels and continua of the excited A 1+u and B 1u

          electronic states.  Theoretical and experimental data are used to character-

          ize the molecular properties taking advantage of knowledge recently obtained

          from photoassociation spectroscopy and ultra-cold atom collision studies. The

          quantum-mechanical calculations are carried out for temperatures in the range

          from 1000 to 2000 K and are compared with previous calculations and mea-

          surements.

          PACS numbers: 33.20.-t, 34.20.Mq, 52.25.Rv



                                                                         Typeset using REVTEX



                                                 1

                                      I. INTRODUCTION



    The absorption spectra of pure alkali-metal vapors at temperatures of the order 1000 K

can be a rich source of information on molecular potentials and properties. Achieving a high

vapor pressure of lithium in experiments requires higher temperatures than the other alkali-

metal atoms, but there are some data from heat pipe ovens [1- 3] and from a specialized

apparatus [4].  In addition to the atomic lines the spectra exhibit gross molecular features

attributable to transitions between bound levels of the ground electronic state and levels of

the excited singlet states and weaker features arising from analogous triplet transitions.

    Theoretically, the envelope of the alkali-metal molecular absorption spectra can be quan-

titatively reproduced using semi-classical models [5- 9] and ro-vibrational structure [7] and

continua [10] can be reproduced from quantum-mechanical models.  In this paper we cal-

culate quantum-mechanically absorption spectra for the X  1+g-A  1+u and X  1+g-B 1u

transitions in Li2. Although both semi-classical [7] and quantum-mechanical [7,11] calcula-

tions have been performed and compared [7] previously for these transitions of Li2, recent

improvements in the molecular data prompt the present comprehensive study.

    From photoassociation spectroscopy and cold collision studies performed in the last few

years as well as recent theoretical work there have been significant critical tests of and

improvements to the molecular potentials [12- 14], particularly at long-range [15- 17], and

transition dipole moment data [15] available for Li2, as well as to the value of the lifetime

of the Li 2p state [18,19]. This paper presents calculations of the spectra over the full range

of wavelengths where absorption in the X  1+g-A  1+u and X  1+g-B 1u bands is possible.

We calculate the satellite feature profiles at various temperatures, identify and explore the

influence of quasibound states and the contributions of bound-bound versus bound-free

transitions, calculate partition functions, and calculate lifetimes for the A  1+u and B 1 u

ro-vibrational levels.



           II. QUANTUM THEORY OF ABSORPTION CROSS SECTION



    In the quantum-mechanical formulation an absorption cross section from a vibration-

rotation state of the lower electronic state (v00; J00; 00) to the vibration-rotation state of the

upper electronic state (v0; J0; 0) is

                                                                             00
                  0J00         8ss3                             2          SJJ0000
               oevv00J0000() = ______|<OEv00J0000|D(R) |OEv0J00>|g( - ij) _________              (1)
                                3hc                                       2J00+ 1



                                                  2
                                                            00
where g( - ij) is a line-shape function of dimension -1 , SJJ0000is the H"onl-London factor

and ij  |Ev0J00 - Ev00J0000| is the transition frequency  [20,21]. In this study, g( - ij) is

approximated by 1=, with  the bin size.  For a bound-free transition, the absorption

cross section from a bound level of the lower electronic state (v00J0000) to a continuum level

of the upper electronic state (ffl0J00) can be written as

                                                                           00
                         0J00         8ss3                             2 SJJ0000
                    oefflv00J0000() = ______|<OEv00J0000|D(R) |OEffl0J00>|_______                (2)
                                       3hc                              2J00+ 1

where the continuum wave function OEffl0J00is energy normalized.  Free-bound or free-free

transitions are not considered since the temperatures studied here are not high enough for

these types of transitions to be important within the singlet manifold.

    The radial wave function can be obtained from the Schr"odinger equation for the relative

motion of the nuclei
                                                                  !
                  d2OE(R)_    2       2           J(J + 1) - 2
                          +   ___2E - ___2V (R) - ________________  OE(R) = 0;                   (3)
                    dR2     h       h                   R2


where V (R) is the rotationless potential energy for the electronic state,  = 6 394:7 is the

reduced mass of the 7Li  atoms, and E is for bound states the eigenvalue EvJ   measured

with respect to the dissociation limit associated with the wave function OE(R) = OEvJ  (R).

Similarly, for continuum states E is the relative kinetic energy of the colliding atoms EfflJ

associated with the energy-normalized wave function OE(R) = OEfflJ(R).

    The total absorption cross section at frequency   can be obtained by averaging over

initial vibration-rotation levels (v00; J00; 00) with a relevant weighting factor and summing

over all possible bound-bound and bound-free transitions with frequencies between  and

 +  [7] yielding
                    "
                        X        0 00
        oe() =  Z-1l          oevvJ00J0000!J00(2J00+ 1) exp [-(De + Ev00J0000)=kT ]
                      J0J00v0v00                                                            #
                        X    Z            0 00
                     +          dffl0oefflvJ00J0000!J00(2J00+ 1) exp [-(De + Ev00J0000)=kT;]     (4)
                       J0J00v00

where !J00is a statistical factor due to nuclear spin with the values [I=(2I + 1)] = 3_8for even

J and [(I + 1)=(2I + 1)] = 5_8for odd J, for 7Li2 with I = 3_2. With the zero of energy taken

to be the potential minimum, the partition function Zl of the lower state with dissociation

energy De is

                            X
                      Zl =      !J00(2J00+ 1) exp [-(De + Ev00J0000)=kT ];                       (5)
                           J00v00

assuming thermodynamic equilibrium.  The resulting cross sections can be used to model

the dimer absorption spectra in the quasistatic limit.



                                                  3

                                   III. MOLECULAR DATA



    The adopted molecular potentials of the X  1+g, A  1+u and B 1u states are shown in

Fig. 1.  The ground X1+g state potential was constructed using the approach of Ref. [22].

We adopted the recommended [23] potential obtained by Barakat et al. [24], who applied

the Rydberg-Klein-Rees (RKR) method to measured energies.  The data were connected

smoothly to the long-range form


                                      C6   C8    C10
                            V (R) = - ___- ___ - ____ + Vexc(R);                                 (6)
                                      R6   R8    R10


where Vexc(R) is the exchange energy [22] and the coefficients C6, C8, and C10 have been

calculated in Refs. [15,25- 27] and we use atomic units in this section.  We adopted the

coefficients C6  =  1 393:39,  C8  =  83 425:8 and C10  =  7:372 x 106  from Ref. [26].   The

two regions were connected at R  =  23:885 a0  yielding a value of 8516.95 cm  -1 for the

dissociation energy of the X1+g state of 7Li2, in satisfactory agreement with the accepted

value of 8516.61 cm  -1 [23].  The form a exp (-bR) was used to extrapolate the potential

at short-range where the constants were determined to smoothly connect to the innermost

RKR points.  The resulting X  1+g potential yields an s-wave scattering length of 33:6 a0,

in excellent agreement with the accepted [28] value of 33  2 and a sensitive test of the

assembled data.

    For the excited A1+u state the RKR potential of Ref. [17] was adopted and smoothly

connected at about R = 140 a0 to the long-range form


                                           C3    C6    C8
                                   V (R) = ___ - ___ - ___;                                      (7)
                                           R3    R6    R8


with coefficients C3 = -11:000 226 and C6 = 2 075:05 from Ref. [26] and C8 = 2:705 x

105 from Ref. [15].  For R  <  3:78 a0 the data were connected to the short range form

a exp (-bR). The dissociation energy for the A1+u state is determined to be 9 352:194 cm -1

in our calculation, in good agreement with the experimental value [29] of 9 352:5(6) cm -1

given in Ref. [30].

    The potential for the B1u state has a hump with maximum at R  11 a0 and various

determinations of the hump location and height have been summarized in Refs. [31] and [30].

We adopted the IPA (Inverted Perturbation Approach) potentials from Hessel and Vidal [32]

for R < 9:35 a0 and the ab initio potential of Schmidt-Mink et al. [30] for 10:5 < R < 30 a0

with one additional point at R = 11:2 a0 fixing the barrier maximum energy at 512 cm -1

above dissociation. At R = 30 a0 the data were connected to the long-range form of Eq. (7)

by shifting down the data by 0:3 cm -1  from 10:5 < R < 30.  The values of the coefficients



                                                  4

used in Eq. (7) were C3 = 5:500 113 and C6 = 1406:08 from Ref. [26] and C8 = 4:756 x 104

from Ref. [15]. The potential energy data for R < 9:35 a0 were fixed using the B 1u state

dissociation energy of 2 984:8 cm -1 , which we determined using the experimental value for

Te of 20 436:32 cm -1  [32], the atomic asymptotic energy of 14 904:0 cm -1 , and the X  1+g

dissociation energy of 8 516:61 cm -1 [23]. Finally, the data in the range 9:35 < R < 10:5 were

smoothly connected using cubic splines.  For R < 4:254 61 a0 the data were extrapolated

using the short range form a exp (-bR).

    Transition dipole moments for the X 1+g-A  1+u and X 1+g-B 1u transitions are avail-

able from ab initio calculations [33,34,30] and for X-A transitions from measured lifetimes of

A state levels [35]. For the electronic transition dipole moments D(R), we adopted ab initio

calculations of Ratcliff, Fish, and Konowalow [33] connected at R = 35 a0 to the long-range

asymptotic form


                                                     b
                                      D1 (R) = D0 + ___:                                         (8)
                                                    R3


The value of the coefficient D0 was 3:3175 for both X-A and X-B transitions and the

coefficient b was 283:07 or -141:53 [15] for, respectively, X-A or X-B transitions. For both

transitions, we multiplied D1 (R) calculated using the above coefficients by a constant such

that the value D1 (35) was identical to the corresponding ab initio  value from Ref. [33]

to provide a smooth connection between short and long-range forms for D(R).  The X-A

dipole moment function that we adopted is consistent with that derived by Baumgartner et

al. [35] from experimental measurements. There is no experimentally-derived dipole moment

function for the X-B transition.



                                          IV. RESULTS



    Bound and continuum wave functions were calculated using the Numerov method to

integrate Eq. (3).  For the X and A states, eigenvalues were generally in good agreement

with the Rydberg-Klein-Rees values used as input to the potentials constructed.  [There

is an apparent misprint for the energy of the v00= 9 level of the X1+g state in Ref. [24].

We used 3 098:641 2 cm -1 , consistent with Ref. [29].] The constructed B 1u state potential

can reproduce the rotationless IPA energies tabulated by Hessel and Vidal typically to

about 0:1 cm -1 , with the greatest discrepancy 0.15 cm -1  for the v0 = 13 value. Calculated

frequencies of B 1u-X  1+gtransitions were also compared with calculations of Verma, Koch,

and Stwalley [36] and the agreement was good, within 0.1 cm -1 . Four quasi-bound levels of



                                                  5

the B 1u state were found.  For the rotationless potential their calculated eigenvalues are

143.91, 276.7, 391.74 and 483.88 cm -1  above the dissociation limit for v0 = 14 to v0 = 17.

    The calculated term energies of vibration-rotation states in the X  1+g state were used

to compute partition functions using Eq. (5).  The maximum vibrational and rotational

quantum numbers in our calculations are 41 and 123,  respectively,  for the X  1+g state.

In the harmonic approximation for ro-vibrational energies, the partition function can be

calculated using the simple expression

                                  X123
                  Z"l  ZR Zv           !J00(2J00+ 1) exp [-hcBeJ00(J00+ 1)=kT ]

                                  J00=0
                                    X41
                                  x     exp [-he(v00+ 1=2)=kT ];                                 (9)
                                    v00=0

which assumes that the term energy can be described by the first terms of the power se-

ries with respect to vibrational and rotational quantum numbers.  Using constants e=c =

351.3904 cm -1  and Be = 0:672 566 cm -1  [32] the partition function from Eq. (9) was calcu-

lated and it is compared with the partition function calculated from Eq. (5) for the X 1+g

state as a function of temperature in Fig. 2. The anharmonicity of the potential for higher

vibrational levels accounts for the differences between the two results with increasing tem-

perature. For J > 2 the X 1+gstate supports quasibound vibrational levels. The expression

Eq. (5) for Zl does not specify whether quasibound states are to be included or not in the

summations.  We evaluated Zl with and without the quasibound levels to ascertain their

importance and the results are shown in Fig. 2. The effect of the additional levels becomes

increasingly significant with higher temperature.  For the present study covering tempera-

tures between 1000 and 2000 K there is not a significant distinction between Zl, Z"l, and the

result with the inclusion of the quasibound states.

    The molecular fraction can be calculated using the expression



                           [NLi2]=[NLi]2 = (QLi2=Q2Li) exp (De=kT );                            (10)



where the atomic partition function QLi is 2(2ssmLikT =h2)3=2, with the electronic partition

function for the atom well-approximated by the spin degeneracy of 2 for the temperatures

studied in the present work, and the molecular partition function QLi2is (2ssmLi2kT =h2)3=2Zl,

with the electronic partition function for the X 1+g state taking the value 1. The molecular

fraction Eq. (10) is plotted in Fig. 2. The absorption coefficient k() can be obtained if the

atomic density is known from



                                      k() = [NLi2]oe():                                         (11)



                                                  6

                                           A. Lifetimes



    Lifetimes of the various ro-vibrational levels of the A 1+u state were measured  [35] and

calculated, see for example, Refs. [37,38,22]. We calculated spontaneous emission transition

probabilities and lifetimes of rotational-vibrational levels of the A1+u state in order to test

the adopted transition dipole moment.  The spontaneous emission rate from a bound state

(v0J00) to a bound state (v00J000) is


                                      64ss43                              2
                  A(v0J00; v00J000) = ________g |<OEv00J000|D(R) |OEv0J00>|;                    (12)
                                        3hc3


where the electronic state degeneracy is


                                          (2 - ffi0;0+00 )
                                      g = _______________                                       (13)
                                             2 - ffi0;0


and we have neglected change in the rotational quantum number.  The total spontaneous

emission rate from the upper level (v0J00) can be obtained by summing over all possible

transitions to bound and continuum states


                         X                       X  Z
             A(v0J00) =      A(v0J00; v00J000) +       dffl00A(v0J00; ffl00J000);               (14)
                        v0000                    00


where A(v0J00; ffl00J000) is the spontaneous emission probability to a continuum energy ffl00

with partial wave J0. The lifetime is



                                       o = 1=A(v0J00):                                          (15)



    Lifetimes of levels v0J0 of the A 1+ustate were measured by Baumgartner et al. [35]. The

A  1+u state is affected by indirect predissociation via the a 3+u  and 1 3+u  states [39,40].

The measured lifetimes om  of vibration-rotation levels thought to be unaffected by indirect

predissociation taken from Ref. [35] and corresponding calculated lifetimes oc are presented

in Fig. 3 along with calculated term energies expressed relative to the potential minimum of

the A state. The energies are plotted in the order of the values listed in Table 1 of Ref. [35]

and correspond to a range of values of (v0; J0) from (0,15) to (24,25). The agreement between

oc and om  is good within the experimental precision of 2 percent [35].  The quasi-bound

levels and continua of the X 1+g state inside or above the high centrifugal potential barriers

are found to be important in calculating lifetimes of high J levels. For instance, by including

transitions to three quasi-bound levels and the continuum states the calculated lifetime of the

A  1+u(v0 = 20; J0 = 50) level changes from 22.7 ns to 19.3 ns which is close to the measured

lifetime of 18:66  0:37 ns [35]. The calculated and measured lifetimes agree well up to about



                                                  7

4 500 cm -1 in agreement with the findings of Ref. [30]. From approximately 5 000 cm -1 and

higher the experimental lifetimes slightly decrease relative to theory by about 0.8 ns or five

percent, as demonstrated in Fig. 3.  We investigated whether the 21+g state might supply

an additional spontaneous decay channel, but the theoretical value Te = 20 128 cm -1 [30] for

this state appears to place its minimum at around 6 000 cm -1 relative to the minimum of the

A  1+u state, apparently ruling this mechanism out. The reason for the significant downturn

of experimental lifetimes for higher term energies is currently not understood; however, the

overall excellent agreement between our calculated lifetimes and the measurements gives

us confidence in our molecular data and calculational procedures.  We compare selected

examples from the present results with calculations by Watson [37] and by Sangfelt et al. [38]

in Table I.   The calculations of Sangfelt et  al.   are larger than ours,  probably because

theoretical transition energies were used. A simple rescaling using experimental energies, as

pointed out by those authors, yields lifetimes in good agreement with the present work. The

dipole moment function calculated by Watson [37] is in good agreement with that adopted

in the present study and cannot account for the shorter lifetimes obtained in that study. We

present a more extensive tabulation of A 1+u lifetimes in Table II covering the same values

tabulated in Table VII of Ref. [38].

    Lifetimes for levels of the B 1u state have been calculated by Uzer, Watson, and Dal-

garno [41],  Uzer and Dalgarno [39],  and Sangfelt et al. [38] and there appear to be no

experimental data.  In Table III we compare our calculated lifetimes of selected B 1u lev-

els with available calculations for higher values of J0.  The present results lie between the

calculations of Uzer et al. [41] and those of Sangfelt et al. [38]. The lifetimes calculated by

Uzer et al. are longer than ours because their transition dipole moment function, calculated

using a model potential method, is smaller than the ab initio  dipole moment of Ratcliff

et al. [33] adopted in the present study. The dipole moment function calculated by Sangfelt

et al., on the other hand, is in good agreement with that adopted in the present work.  As

those authors pointed out, and as our results illustrate, the utilization in their calculation of

calculated excitation energies which were larger than experimental energies yielded lifetimes

that were too short. In Table IV we present a more extensive tabulation of B 1u lifetimes

covering the same values tabulated in Table VIII of Ref. [38].



                                  B. Absorption cross sections



    Absorption spectra arising from molecular singlet transitions at the far wings of the

atomic 2p line at 671 nm consist of a blue wing due to X  1+g-B 1u  transitions and a



                                                  8

red wing due to X 1+g-A  1+u transitions.  Calculations for bound-bound (bb) and bound-

free (bf) absorption cross sections were carried out separately using Eq. (4) with   =

8 cm  -1.   The results for the total (the sum of bb and bf) absorption cross sections at

temperatures of 1000 K, 1500 K and 2033 K are given in Figs. 4-6.  The ratios of the

peak cross sections of the X 1+g-B 1u wing to those of the X 1+g-A  1+u wing are higher

at lower temperatures.  As temperature increases, absorption spectra spread out from the

peak spectral regions and there emerges near 900 nm a satellite feature arising from the

minimum [42] in the X 1 +g-A  1+u difference potential and the maximum of the transition

dipole moment function [43,33].  We also show the bf contributions to the cross sections

on each plot.  The bf component contributes mainly to the extreme blue part of the blue

wing and increases in magnitude significantly as the temperature increases. It is found that

transitions to quasibound and continuum levels of the B 1u state contribute significantly

to the total absorption spectra in the case of X  1+g-B 1u transitions, apparently because

there is less vibrational oscillator strength density in the discrete part of the spectrum for

the B state compared to the A state.  Transitions into quasibound states of the A  1+u or

B 1u states have been included in the results for the total cross sections in Figs. 4-6.

    Theoretical  quantum-mechanical  calculations  for  absorption  cross  sections  from  the

X  1+g state to the A  1+u state over the spectral range 600-950 nm were carried out by

Lam et al. [7] using a constant dipole moment of 6.58 D at 1020 K, for which bf transitions

are not significant.  Our calculations in Fig. 4 are about a factor of 10 less than the result

shown in Fig. 5 of Ref. [7], but agree well both in overall shape and in details of finer struc-

tures.  We repeated the calculations using the constant dipole moment of Ref. [7] for both

classical and quantum-mechanical cross sections and although these two results agreed with

each other, they were also a factor of 10 less than the result shown in Fig. 5 of Ref. [7].

Thus it appears to us that there may be a mislabeling of the vertical axis in Fig. 5 of Lam et

al. [A similar calculation that we performed [44] for Na 2 at 800 K is in complete agreement

with Fig. 4 of Ref. [7].]

    Calculations of absorption spectra at 2033 K over the spectral range 450-750 nm pre-

sented in Fig. 6 are in good agreement with the measured values of Erdman et al. [4] and

with quantum-mechanical calculations performed by Mills [11].  The experimental study of

Erdman et al. [4] involved an investigation of molecular triplet states [4] and did not explore

the satellite feature at 900 nm.  The calculations over the range 450-750 nm by Mills [11]

included triplet molecular transitions and are not directly comparable with the present re-

sults.  Nevertheless since the singlet transitions dominate the absorption we find excellent

qualitative agreement with the calculations presented by Mills.



                                                  9

                                   ACKNOWLEDGMENTS



    We thank R. C^ote for generously sharing assembled data and A. Dalgarno for helpful

discussions.  We also are grateful to A. Gallagher, W. Stwalley, and M. Fajardo for helpful

correspondence. This work is supported in part by the National Science Foundation under

grant PHY97-24713 and by a grant to the Institute for Theoretical Atomic and Molecular

Physics at Harvard College Observatory and the Smithsonian Astrophysical Observatory.



                                                 10

                                             TABLES



_TABLE_I._Comparison_of_calculated_lifetimes_in_ns_for_ro-vibrational_levels_of_the_A_1+u_state.___*
 *__
___________________________________________________________________________________________________*
 *__

_v0__________J0___________Watson_[37]____________Sangfeldt_et_al._[38]__________This_work__________*
 *__

 7           15               16.8                      19.25                     18.55

 9            5               16.9                      19.29                     18.65

20            8               17.3                                                19.04
___________________________________________________________________________________________________*
 *___



    TABLE II. Lifetimes in ns for ro-vibrational levels of the A 1+u state calculated as described

in_the_text._______________________________________________________________________________________*
 *___

v0__________________________J0_=_0__________________________J0_=_9___________________________J0_=_1*
 *5_

 0                           17.74                           17.77                            17.82

 1                           17.87                           17.90                            17.94

 2                           17.98                           18.01                            18.06

 3                           18.09                           18.12                            18.17

 4                           18.20                           18.23                            18.27

 5                           18.30                           18.33                            18.37

 6                           18.39                           18.42                            18.46

 7                           18.48                           18.51                            18.55
___________________________________________________________________________________________________*
 *___



                                                 11

TABLE_III._Comparison_of_calculated_lifetimes_in_ns_for_ro-vibrational_levels_of_the_B_1u_state.___*
 *__
___________________________________________________________________________________________________*
 *__

v0__________J0___________Uzer_et_al._[41]_________Sangfeldt_et_al._[38]_________This_work__________*
 *__

 0          15                 8.3                        6.83                     7.66

 5           9                 8.5                        7.20                     7.95
___________________________________________________________________________________________________*
 *___



    TABLE IV.  Lifetimes in ns for vibrational-rotational levels of the B 1u state calculated as

described_in_the_text._____________________________________________________________________________*
 *___

v0__________________________J0_=_1__________________________J0_=_9___________________________J0_=_1*
 *5_

 0                           7.65                            7.65                             7.66

 1                           7.70                            7.71                             7.72

 2                           7.76                            7.76                             7.77

 3                           7.81                            7.82                             7.83

 4                           7.88                            7.89                             7.90

 5                           7.94                            7.95                             7.97

 6                           8.02                            8.03                             8.04

 7                           8.10                            8.10                             8.12
___________________________________________________________________________________________________*
 *___



                                                 12

                                   FIGURES



FIG. 1. Adopted potentials for the X 1+g, A 1+u, and B 1u electronic states.



                                       13




    FIG. 2.  Comparison at various temperatures of the partition functions Z"l, calculated using

Eq. (9) and experimentally determined spectroscopic constants, curve A, and Zl, from Eq. (5)

and numerically determined eigenvalues, curves B and C. Inclusion of quasibound states in the

calculation of Zl results in curve C as discussed in the text.  Curve D represents the molecular

fraction [NLi2]=[NLi]2, Eq. (10), as a function of temperature.



                                                 14




    FIG. 3.  Comparison of our calculated lifetimes (diamonds) and measured lifetimes (circles)

from Ref. [35].  The error bars indicate the quoted experimental uncertainty of 2 percent.  The

levels given are those that were measured, ordered by increasing energy, as listed in Table 1 of

Ref. [35].



                                                 15




    FIG. 4.  Total absorption cross sections from X 1+g-A 1+u and X 1+g-B 1u transitions in-

cluding bound to bound and bound to free transitions at a temperature of 1000 K. The satellite

feature near 900 nm does not appear at this temperature and bound-free absorption (dotted curve)

is insignificant. The inset presents a magnified view of the bound-free contribution.



                                                 16




    FIG. 5. Total absorption cross sections from X 1+g-A 1+uand X 1+g-B 1u transitions includ-

ing bound to bound and bound to free transitions at a temperature of 1500 K. As the temperature

increases, the absorption spectra are distributed over a wider spectral range and the ratio of the

peak cross sections between X 1+g-B 1u and X 1+g-A 1+ubands decreases. The satellite feature

near 900 nm and bound-free absorption (dotted curve) are noticeable at this temperature.  The

inset presents a magnified view of the bound-free contribution.



                                                 17




    FIG. 6. Total absorption cross sections from X 1+g-A 1+uand X 1+g-B 1u transitions includ-

ing bound to bound and bound to free transitions at a temperature of 2033 K. At this temperature

the satellite feature near 900 nm is now prominent and bound-free absorption (dotted curve) con-

tributes significantly at 450 nm. The inset presents a magnified view of the bound-free contributio*
 *n.



                                                 18

                                        REFERENCES



  [1]C. R. Vidal, J. Appl. Phys. 44, 2225 (1973).


  [2]D. Veza, S. Milosevic, and G. Pichler, Chem. Phys. Lett. 93, 401 (1982).


  [3]W. Theiss, H. J. M"uschenborn, and W. Demtr"oder, Chem. Phys. Lett. 174, 126 (1990).


  [4]P. S. Erdman et al., Chem. Phys. Lett. 252, 248 (1996).


  [5]D. R. Bates, Mon. Not. R. Astron. Soc. 112, 40 (1952).


  [6]J. Szudy and W. E. Baylis, J. Quant. Spectr. Rad. Trans. 15, 641 (1975).


  [7]L. K. Lam, A. Gallagher, and M. M. Hessel, J. Chem. Phys. 66, 3550 (1977).


  [8]J. Schlejen, J. P. Woerdman, and G. Pichler, J. Mol. Spect. 128, 1 (1988).


  [9]J. P. Woerdman et al., J. Phys. B 18, 4205 (1985).


[10] R. A. Buckingham, S. Reid, and R. Spence, Mon. Not. R. Astron. Soc. 112, 382 (1952).


[11] J. D. Mills, Technical Report No. PL-TR-94-3001, Propulsion Directorate, Phillips Lab-

     oratory, Air Force Materiel Command (unpublished).


[12] R. C^ote and A. Dalgarno, Phys. Rev. A 50, 4827 (1994).


[13] E. R. I. Abraham, W. I. McAlexander, C. A. Sackett, and R. G. Hulet, Phys. Rev. Lett.

     74, 1315 (1995).


[14] E. R. I. Abraham et al., Phys. Rev. A. 53, R3713 (1996).


[15] M. Marinescu and A. Dalgarno, Phys. Rev. A 52, 311 (1995).


[16] Z.-C. Yan, A. Dalgarno, and J. F. Babb, Phys. Rev. A 55, 2882 (1997).


[17] F. Martin et al., Phys. Rev. A 55, 3458 (1997).


[18] Z.-C. Yan and G. W. F. Drake, Phys. Rev. A 52, R4316 (1995).


[19] E. R. I. Abraham et al., Phys. Rev. A. 54, R5 (1996).


[20] H. Friedrich, Theoretical Atomic Physics (Springer, Berlin, 1991).


[21] H. Lefebvre-Brion and R. W. Field, Perturbations in the spectra of diatomic molecules

     (Academic, Orlando, 1986).


[22] R. C^ote and A. Dalgarno, Phys. Rev. A 58, 498 (1998).



                                                 19

[23] W. T. Zemke and W. C. Stwalley, J. Phys. Chem. 97, 2053 (1993).


[24] B. Barakat et al., Chem. Phys. 102, 215 (1986).


[25] B. Bussery and M. Aubert-Frecon, J. Chem. Phys. 82, 3224 (1985).


[26] Z.-C. Yan, J. F. Babb, A. Dalgarno, and G. W. F. Drake, Phys. Rev. A 54, 2824 (1996).


[27] M. Rerat, B. Bussery, and M. Frecon, J. Mol. Spect. 182, 260 (1997).


[28] J. Weiner, V. S. Bagnato, S. Zilio, and P. S. Julienne, Rev. Mod. Phys. 71, 1 (1999).


[29] P. Kusch and M. M. Hessel, J. Chem. Phys. 67, 586 (1977).


[30] I. Schmidt-Mink, W. Mueller, and W. Meyer, Chem. Phys. 92, 263 (1985).


[31] M. L. Olson and D. D. Konowalow, Chem. Phys. 22, 29 (1977).


[32] M. M. Hessel and C. R. Vidal, J. Chem. Phys. 70, 4439 (1979).


[33] L. B. Ratcliff, J. L. Fish, and D. D. Konowalow, J. Molec. Spectrosc. 122, 293 (1987).


[34] D. D. Konowalow, M. E. Rosenkrantz, and D. S. Hochhauser, J. Mol. Spect. 99, 321

     (1983).


[35] G. Baumgartner, H. Kornmeier, and W. Preuss, Chem. Phys. Lett. 107, 13 (1984).


[36] K. K. Verma, M. E. Koch, and W. C. Stwalley, J. Chem. Phys. 78, 3614 (1983).


[37] D. K. Watson, Chem. Phys. Lett. 51, 513 (1977).


[38] E. Sangfelt, H. A. Kurtz, N. Elander, and O. Goscinski, J. Chem. Phys. 81, 3976 (1984).


[39] T. Uzer and A. Dalgarno, Chem. Phys. 51, 271 (1980).


[40] I. Schmidt-Mink and W. Meyer, Chem. Phys. Lett. 121, 49 (1985).


[41] T. Uzer, D. K. Watson, and A. Dalgarno, Chem. Phys. Lett. 55, 6 (1978).


[42] A. Jablonski, Phys. Rev. 68, 78 (1945).


[43] J. P. Woerdman, J. Chem. Phys. 75, 5577 (1981).


[44] H.-K. Chung, K. Kirby, and J. F. Babb, in preparation (unpublished).



                                                 20