density functional calculations on low-lying singly excited states of open-shell atoms

7
13 November 1998 Ž . Chemical Physics Letters 296 1998 530–536 Density functional calculations on low-lying singly excited states of open-shell atoms Ranbir Singh 1 , Amlan K. Roy, B.M. Deb ),2 Theoretical Chemistry Group, Department of Chemistry, Panjab UniÕersity, Chandigarh-160014, India Received 8 June 1998; in final form 15 September 1998 Abstract By employing a simple density-functional approach, low-lying singly excited states of several open-shell atoms, viz. B, C, O, F, Na, Mg, Al, Si, P and Cl, have been calculated and compared with experimental results. The work-function-based exchange potential has been used in a nonrelativistic, single-determinantal framework. The effects of two different Ž . correlation energy functionals local Wigner and nonlocal Lee–Yang–Parr on excitation energies and excited-state energies have been studied. While the exchange-only results show good agreement with numerical Hartree–Fock results, the correlation energy functionals do not show any significant improvements in excitation energies over the exchange-only results, although the excited-state energies are improved significantly. q 1998 Elsevier Science B.V. All rights reserved. 1. Introduction Ž . Density functional theory DFT has achieved impressive successes in explaining the electronic structure and properties of many-electron systems in their ground state. However, the calculation of ex- cited-state energies and densities has remained a bottleneck. Recently, a simple DFT-based approach has been employed for dealing with atomic singly, ) Corresponding author. Fax: q91-0172-541409; telex: 395- 7464 RSIC IN; e-mail: [email protected] 1 Present Address: Department of Physics, Brooklyn College of the City University of New York, Brooklyn, NY 11210, USA. 2 Also from the Jawaharlal Nehru Centre for Advanced Scien- tific Research, Bangalore-560064, India. doubly, triply, valence- and core-excited states of w atoms, including autoionizing and satellite states 1– x 7 . Excited-state energies, radial densities, excitation Ž . energies 8–2000 eV and other energy differences Ž . 0.03–23.5 eV were satisfactorily reproduced within a single-determinantal framework. Ž. The objectives of this paper are: i to calculate several low-lying singly excited states of open-shell Ž . atoms B, C, O, F, Na, Mg, Al, Si, P and Cl through Ž. the present single-determinantal approach; ii to study the effects of two different correlation energy Ž . functionals, viz. Wigner W and Lee–Yang–Parr WC Ž W , both of which are designed for ground-state LYP . Ž . calculations , on these excited states; and iii to find the efficacy of the present approach, using spherical densities, especially in the light of large differences between calculated DFT and experimental excitation wx energies, as reported by Nagy and Andrejkovics 8 . 0009-2614r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved. Ž . PII: S0009-2614 98 01031-8

Upload: ranbir-singh

Post on 02-Jul-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Density functional calculations on low-lying singly excited states of open-shell atoms

13 November 1998

Ž .Chemical Physics Letters 296 1998 530–536

Density functional calculations on low-lying singly excited statesof open-shell atoms

Ranbir Singh 1, Amlan K. Roy, B.M. Deb ) ,2

Theoretical Chemistry Group, Department of Chemistry, Panjab UniÕersity, Chandigarh-160014, India

Received 8 June 1998; in final form 15 September 1998

Abstract

By employing a simple density-functional approach, low-lying singly excited states of several open-shell atoms, viz. B,C, O, F, Na, Mg, Al, Si, P and Cl, have been calculated and compared with experimental results. The work-function-basedexchange potential has been used in a nonrelativistic, single-determinantal framework. The effects of two different

Ž .correlation energy functionals local Wigner and nonlocal Lee–Yang–Parr on excitation energies and excited-state energieshave been studied. While the exchange-only results show good agreement with numerical Hartree–Fock results, thecorrelation energy functionals do not show any significant improvements in excitation energies over the exchange-onlyresults, although the excited-state energies are improved significantly. q 1998 Elsevier Science B.V. All rights reserved.

1. Introduction

Ž .Density functional theory DFT has achievedimpressive successes in explaining the electronicstructure and properties of many-electron systems intheir ground state. However, the calculation of ex-cited-state energies and densities has remained abottleneck. Recently, a simple DFT-based approachhas been employed for dealing with atomic singly,

) Corresponding author. Fax: q91-0172-541409; telex: 395-7464 RSIC IN; e-mail: [email protected]

1 Present Address: Department of Physics, Brooklyn College ofthe City University of New York, Brooklyn, NY 11210, USA.

2 Also from the Jawaharlal Nehru Centre for Advanced Scien-tific Research, Bangalore-560064, India.

doubly, triply, valence- and core-excited states ofwatoms, including autoionizing and satellite states 1–

x7 . Excited-state energies, radial densities, excitationŽ .energies 8–2000 eV and other energy differences

Ž .0.03–23.5 eV were satisfactorily reproduced withina single-determinantal framework.

Ž .The objectives of this paper are: i to calculateseveral low-lying singly excited states of open-shell

Ž .atoms B, C, O, F, Na, Mg, Al, Si, P and Cl throughŽ .the present single-determinantal approach; ii to

study the effects of two different correlation energyŽ .functionals, viz. Wigner W and Lee–Yang–ParrWC

Ž . ŽW , both of which are designed for ground-stateLYP. Ž .calculations , on these excited states; and iii to find

the efficacy of the present approach, using sphericaldensities, especially in the light of large differencesbetween calculated DFT and experimental excitation

w xenergies, as reported by Nagy and Andrejkovics 8 .

0009-2614r98r$ - see front matter q 1998 Elsevier Science B.V. All rights reserved.Ž .PII: S0009-2614 98 01031-8

Page 2: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536 531

2. Method of calculation

The work-function-based nonvariational exchangew xpotential is 9,10

r ™™ ™ ™W r sy e r d l 1Ž . Ž . Ž .Hx x`

™ ™Ž .where e r is the electric field due to the Fermi-holex™™XŽ .charge distribution r r,r and l denotes the pathx

of integration. The two correlation potentials em-ployed are:

Ž .i The local parametrized Wigner-type functionalw x11

y1r3aqbr™W r sy 2Ž . Ž .WC 2y1r3aqcrŽ .

where as9.81, bs28.583 and cs21.437.Ž .ii The closed-shell, nonlocal functional of Lee etw xal. 12

™ X 5r3W r sya F rqF yabC rŽ . Ž .LYP 1 1 F

=8 ab

2X Y < <G rq G y G r =r1 1 1ž /3 42X 2 2< <qG 3=r q2 r= r q4G = rŽ .1 1

ab2 2Y X< < < <y 3G r =r qG 5=rŽ1 172

2 2q6r= r q4G = r 3Ž .. 1

Ž . Ž y 1r 3 .y 1 Ž .where F r s 1 q d r ; G r s1 1Ž . y5r3 Ž y1r3.F r r exp ycr , as0.04918, bs0.132,1

Ž 2 .2r3 Xcs0.2533, ds0.349, C s3 3p r10. F andF 1

GX are the first derivatives respectively, with respect1

to r; GY is the second derivative.1

With these potentials, a nonrelativistic Kohn–Sham-type differential equation is solved numeri-cally in the central-field approximation,

1™ ™ ™ ™2y = qW r qW r f r se f rŽ . Ž . Ž . Ž .es xc i i i2

4Ž .in order to obtain the self-consistent set of orbitals

™� 4 Ž .f . W r is the Hartree electrostatic potential,i es

including electron-nuclear attraction and interelec-™Ž .tronic repulsion; W r is the exchange-correlationxc

potential, W qW , where W is either W orx c c WC

W .LYP

Ž .The total energy is the sum of kinetic T , electro-Ž . Ž .static V and exchange-correlation V energies:es xc

1™ ™ ™) 2Tsy f r = f r d r , 5Ž . Ž . Ž .ÝH i i2 i

™ ™ ™Xr r 1 r r r rŽ . Ž . Ž .

X™ ™ ™V syz d rq d r d r ,H HHes X™ ™r 2 < <ryr6Ž .

™ ™™X1 r r r r ,rŽ . Ž .x X™ ™V s d r d r ;HHx X™ ™2 < <ryr2X™™

g r ,rŽ .X™™

r r ,r sy , 7Ž .Ž .x ™2 r rŽ .V sV qV ; V sV or V , 8Ž .xc x c c WC LYP

™r rŽ .

™V sy d r , 9Ž .HWC y1r39.81q21.437r

1y2r3 5r3V sya rqbr C rHLYP Fy1r3 ½1qd r

1 12y2 t q t q = rw W 5ž /9 18

= ™y1r3exp ycr d r , 10Ž .Ž .2™1 =r r 1Ž .

™2t s y = r r . 11Ž . Ž .W ™8 8r rŽ .™� Ž .4The orbitals f r are used to construct variousi

determinants which in turn can be employed tocalculate the various multiplets associated with aparticular electronic configuration. The use of Slater’s

w xdiagonal sum rule 13 for calculating the multipletsw xhas been described earlier 1–7 . This procedure was

w xalso employed by others 14–19 .

3. Results and discussion

Table 1 reports the ground-state, excited-state andexcitation energies of the atoms studied, using W -x

only, W qW and W qW , comparing themx WC x LYPŽ . w xwith numerical Hartree–Fock HF 20 , other DFT

w x w x8 and experimental 23 results. For each atom,calculations were performed for an excited statecorresponding to an electronic configuration where

Page 3: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536532

Table 1Nonrelativistic total energies and excitation energies from the ground state, in a.u., of open-shell atomic states. ‘‘Exact’’ ground-stateenergies are given in parentheses in column 4. For every state, the top number in columns 4 and 6 is the W -only result whereas the middlex

number and the third number refer to W qW and W qW respectively. Except in column 4, all numbers in parentheses denotex WC x LYP

percentage deviations with respect to experimental results. In column 7, CA:VWN denotes Ceperley–Alder: Vosko–Wilk–Nusair

Ž .Atom state Total energy yE Excitation energya a c dHF Present HF Present CA:VWN Expt.

22B 2s 2p P 24.5290 24.528924.652024.6616

bŽ .24.65442 Ž .B 2s2p P 24.4507 24.4491 0.0783 0.0798 y39.2 0.209 0.1312

Ž . Ž . Ž .24.5722 y40.3 0.0798 y39.2 59.3Ž .24.5831 0.0785 y40.2

2 Ž .D 24.3119 24.3108 0.2171 0.2181 0.0 0.2181Ž . Ž .24.4324 y0.46 0.2196 0.69

Ž .24.4432 0.2184 0.142 Ž .S 24.2481 24.2347 0.2809 0.2942 1.59 0.2896

Ž . Ž .24.3553 y3.00 0.2967 2.45Ž .24.3661 0.2955 2.04

2 Ž .P 24.1790 24.1547 0.3500 0.3742 13.2 0.3305Ž . Ž .24.2752 5.90 0.3768 14.0

Ž .24.2863 0.3753 13.632 2C 2s 2p P 37.6886 37.6881

37.848437.8647

bŽ .37.84553 Ž .C 2s2p S 37.5992 37.5972 0.0894 0.0909 y40.9 0.302 0.1537

Ž . Ž . Ž .37.7579 y41.8 0.0905 y41.1 96.5fŽ .37.7753 0.0894 y41.8 0.1152

Ž .y25.03 fŽ .D 37.3944 37.3936 0.2942 0.2945 0.86 0.3087 0.2920

Ž . Ž . Ž .37.5525 0.75 0.2959 1.34 5.72Ž .37.5702 0.2945 0.86

3 fŽ .P 37.3370 37.3640 0.3516 0.3241 y5.48 0.3662 0.3429Ž . Ž . Ž .37.5229 2.54 0.3255 y5.07 6.79

Ž .37.5404 0.3243 y5.421 fŽ .D 37.1696 37.1638 0.5190 0.5243 17.6 0.5389 0.4460

Ž . Ž . Ž .37.3209 16.4 0.5275 18.3 20.8Ž .37.3390 0.5257 17.9

3 fŽ .S 37.1421 37.0764 0.5465 0.6117 26.9 0.5648 0.4821Ž . Ž . Ž .37.2337 13.4 0.6147 27.5 17.2

Ž .37.2522 0.6125 27.11 fŽ .P 37.1158 37.0742 0.5728 0.6139 12.4 0.5949 0.5462

Ž . Ž . Ž .37.2305 4.87 0.6179 13.1 8.92Ž .37.2484 0.6163 12.8

32 4O 2s 2p P 74.8094 74.808875.058375.0838

bŽ .75.06735 Ž .O 2s2p P 74.1839 74.1826 0.6255 0.6262 8.85 0.5753

Ž . Ž .74.4303 8.73 0.6280 9.16Ž .74.4574 0.6264 8.88

1 Ž .P 73.8720 73.8682 0.9374 0.9406 8.74 0.8650Ž . Ž .74.1145 8.37 0.9438 9.11

Page 4: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536 533

Ž .Table 1 continued

Ž .Atom state Total energy yE Excitation energya a c dHF Present HF Present CA:VWN Expt.

Ž .74.1422 0.9416 8.8622 5F 2s 2p P 99.4093 99.4136

99.714699.7418

bŽ .99.73426 Ž .F 2s2p S 98.5312 98.5357 0.8781 0.8779 14.3 0.7680

Ž . Ž .98.8346 14.3 0.8800 14.6Ž .98.8636 0.8782 14.4

2Na 3s S 161.8589 161.8669162.2634162.2856

bŽ .162.2572 Ž .Na 3p P 161.7864 161.7946 0.0725 0.0723 y6.47 0.0775 0.0773

Ž . Ž . Ž .162.1891 y6.21 0.0743 y3.88 0.26Ž .162.2058 0.0798 3.23

12Mg 3s S 199.6146 199.6281200.0658200.0882

bŽ .200.0593 Ž .Mg 3s3p P 199.5467 199.5593 0.0679 0.0688 30.9 0.0995 0.0996

Ž . Ž . Ž .199.9954 y31.8 0.0704 y29.3 y0.10Ž .200.0155 0.0727 y27.0

1 Ž .P 199.4711 199.4659 0.1435 0.1622 1.57 0.1597Ž . Ž .199.8874 y10.1 0.1784 11.7

Ž .199.9061 0.1821 14.022Al 3s 3p P 241.8767 241.8943

242.3733242.3954

bŽ .242.35642 Ž .Al 3s3p P 241.7909 241.8067 0.0858 0.0876 y33.7 0.1322

Ž . Ž .242.2848 y35.1 0.0885 y33.1Ž .242.3055 0.0899 y32.0

2 D 241.6917 241.7081 0.1850 0.1862242.1849 0.1884242.2051 0.1903

2 Ž .S 241.6453 241.6529 0.2314 0.2414 2.37 0.2358Ž . Ž .242.1290 y1.87 0.2443 3.60

Ž .242.1491 0.2463 4.4522 Ž .Al 3s3p P 241.6061 241.5949 0.2706 0.2994 16.0 0.2581

Ž . Ž .242.0705 4.84 0.3028 17.3Ž .242.0904 0.3050 18.2

32 2Si 3s 3p P 288.8544 288.8746289.3977289.4211Ž .b289.374

53 eŽ .Si 3s3p S 288.7630 288.7813 0.0914 0.0933 y38.2 0.247 0.151Ž . Ž . Ž .289.3038 y39.5 0.0939 y37.8 63.6

Ž .289.3262 0.0949 y37.23 Ž .D 288.6201 288.6398 0.2343 0.2348 6.49 0.2205

Ž . Ž .289.1610 6.26 0.2367 7.35Ž .289.1832 0.2379 7.89

3P 288.5803 288.5988 0.2741 0.2758

Page 5: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536534

Ž .Table 1 continued

Ž .Atom state Total energy yE Excitation energya a c dHF Present HF Present CA:VWN Expt.

289.1194 0.2783289.1416 0.2795

1D 288.4664 288.4796 0.3880 0.3950288.9990 0.3987289.0212 0.3999

3S 288.4493 288.4371 0.4051 0.4375288.9567 0.4410288.9790 0.4421

1P 288.4292 288.4374 0.4252 0.4372288.9570 0.4407288.9784 0.4427

42 3P 3s 3p S 340.7188 340.7405341.3103341.3352

bŽ .341.27244 Ž .P 3s3p P 340.4165 340.4379 0.3023 0.3026 11.5 0.2713

Ž . Ž .341.0057 11.4 0.3046 12.3Ž .341.0301 0.3051 12.5

24 Ž .P 3s3p D 340.2776 340.2997 0.4412 0.4408 33.7 0.3297Ž . Ž .340.8663 33.8 0.4440 34.7

Ž .340.8908 0.4444 34.82 Ž .S 340.2093 340.2267 0.5095 0.5138 54.6 0.3323

Ž . Ž .340.7930 53.3 0.5173 55.7Ž .340.8173 0.5179 55.9

2 Ž .P 340.1430 340.1528 0.5758 0.5877 90.0 0.3094Ž . Ž .340.7184 86.1 0.5919 91.3

Ž .340.7429 0.5923 91.422 5Cl 3s 3p P 459.4821 459.5061

460.1749460.2030

bŽ .460.19626Cl 3s3p S 458.9168 458.9407 0.5653 0.5654

459.6072 0.5677459.6355 0.5675

a w xNumerical Hartree–Fock results, taken from Ref. 20 .b w x w xGround-state results, taken from Ref. 21 for ZF10 and from Ref. 22 for Z)10 with Lamb shift correction.c w xOther DFT results, taken from Ref. 8 . See also footnote f below.d w x Ž 3 5 .Taken from Ref. 23 , except for Si 3s3p , S .e w xAs cited in Ref. 8 .f Ž . w xResults from 48-state calculation theory 2 , from Ref. 24 .

an electron in the outermost s orbital was excited tothe adjacent p orbital. For both ground and excitedstates considered here, the W -only results agreex

quite well with HF results, with deviations in therange of 0.0004–0.2%. Inclusion of correlation leadsto a significant improvement in the ground-state

w xresults, compared with the ‘‘exact’’ results 21,22 ,with deviations in the range of 0.002–0.01%. There-fore, one might assume that the excited-state ener-

gies would also improve significantly due to correla-tion, leading to improved estimates of the excitationenergies relative to the ground state. Interestingly,the second part of this assumption is not borne outby the results in Table 1, when compared with

Ž 3 3 .experimental results; except for C 2s2p , P , NaŽ 2 . Ž 3 . Ž 2 4 .3p, P , Mg 3s3p, P , Al 3s3p , P and SiŽ 3 5 .3s3p , S , the W -only excitation energies are bet-x

ter than those obtained by using correlation. In other

Page 6: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536 535

words, if only excitation energies are required then,due to a partial cancellation of errors, even the HFŽ .or exchange-only DFT results generally give betterestimates of excitation energies for these atoms.Furthermore, both the local Wigner correlation func-tional and the nonlocal LYP functional yieldground-state energies in good agreement with the‘‘exact’’ results.

Ž 2 4 .For the five states in Table 1, viz. B 2s2p , P ,Ž 3 5 . Ž 2 . Ž 3 .C 2s2p , S , Na 3p, P , Mg 3s3p, P and Si

Ž 3 5 . w x3s3p , S , other DFT results 8 show deviationsfrom experimental results in the range of 0.1–96.5%,compared with the present deviations of 3.2–41.8%.In general, wherever a comparison with experimentalresults could be made, the present calculated results

Ž .show large deviations as high as 91% making themunsuitable for use as a guideline for the experimen-talist. This is indeed puzzling in view of the earlier

w xsuccesses of the same approach 1–7 and the factthat even though open-shell atoms are being studied,the states investigated are singly excited andlow-lying. It is tempting to argue that such failures

Ž .are due to i inherent ‘‘weaknesses’’ of DFT inŽ .dealing with excited states, ii the limitations of the

present single determinantal approach in dealing withŽopen-shell systems i.e. not representing an excited

state as a linear combination of a fairly large numberof wavefunctions of the same space and spin symme-

. Ž . Žtry , iii the present fully numerical basis-set-inde-.pendent calculations apparently not including con-

Ž .tinuum functions, and iv the nonuniversality ofWigner and LYP ground-state correlation functionalswith regard to all states. However, these reasoningsdo not appear to be tenable, especially in view of therecent calculations of excited states of the carbon

w xatom by Dunseath et al. 24 . Their results on C3 Ž5 3 3 1 3 1 .2s2p S, D, P, D, S and P are included in

Table 1 for comparison. In perhaps the most elabo-rate calculations on these states reported so far, theseauthors employ an R-matrix method, with continuumorbitals and 48 states to represent each of the abovesix terms. Their results deviate from experiment by5.7–25.0%, compared with present deviations of0.9–41.8%. As yet, no viable explanation for suchlarge deviations has been forthcoming.

One might also feel that the present discrepanciesmight be due to the assumption of spherical symme-

™Ž .try in calculating W r . In other words, the rota-x

™Ž .tional component in W r , which was negligible inxw xall the excited states studied previously 1–7 , may

now play a significant role in the excited statesŽstudied in this paper note that, for Ne satellites, both

w xLDA and Becke 25 exchange potentials yieldedpoor excitation energies compared to the work-func-

w x.tion exchange potential 7 . However, this is notsupported by the present results in Table 1. For the

3 Ž5 .worst case with C 2s2p S , the exchange energy isin error by 0.002 a.u. whereas for the next worst caseŽ3 . Ž3 .S , the error is 0.066 a.u.; the best case D showsan error of 0.001 a.u. For all the other states in Table1, the error in exchange energy is from 0.001–0.041

4 Ž2 2 .a.u. Still, P 3s3p S, P , with errors in excitationenergy of 56 and 91%, respectively, show only 0.017and 0.009 a.u. errors, respectively, in exchange en-ergy. Clearly, exchange energy is not the reason forsuch large discrepancies between the calculated andexperimental excitation energies; these arise due tothe inability of the present single-determinantal ap-proach to describe electron correlation satisfactorilyin the present excited states which might requiresignificant mixing of ‘‘doubly-excited’’ determinantsfor their proper description. It would be of interest to

wapply multireference coupled-cluster methods 26–x28 as well as other density-functional approaches

Ž .such as the time-dependent TD optimized effectivew xpotential 29 and the TD density functional response

w xtheory 30–32 to the present excited states so thatthe nature of their electron correlation may be better

Ž .understood. In particular, the response theory RTapproach is a purely density-based method in con-trast to the present method which uses both density-based and wavefunction-based approaches to de-

Ž w xscribe the atomic excited states see also Ref. 33 fora combination of density functional and configura-

.tion interaction methods for molecular excited states .The RT approach calculates the linear responseŽ .first-order change in density of the system to a TD

Žperturbation, leading to the dynamic frequency-de-.pendent dipole polarizability whose poles and

residues yield the excitation energies and oscillatorstrengths, respectively. The method has been suc-cessfully applied to calculate molecular excitationenergies below 20 eV for N , CO, CH O, C H ,2 2 2 4

w x Ž w xpyridine and porphin 30–32 see Ref. 34 for areview on density functional approaches to excited

.states .

Page 7: Density functional calculations on low-lying singly excited states of open-shell atoms

( )R. Singh et al.rChemical Physics Letters 296 1998 530–536536

Acknowledgements

We thank the Council of Scientific and IndustrialResearch, New Delhi, University Grants Commis-sion. New Delhi, and the Jawaharlal Nehru Centrefor Advanced Scientific Research, Bangalore, forfinancial support.

References

w x Ž .1 R. Singh, B.M. Deb, Proc. Ind. Acad. Sci. Chem. Sci. 106Ž .1994 1321.

w x Ž . Ž .2 R. Singh, B.M. Deb, J. Mol. Struc. Theochem 361 199633.

w x Ž .3 R. Singh, B.M. Deb, J. Chem. Phys. 104 1996 5892.w x Ž .4 A.K. Roy, R. Singh, B.M. Deb, J. Phys. B 30 1997 4763.w x5 A.K. Roy, R. Singh, B.M. Deb, Int. J. Quant. Chem. 65

Ž .1997 317.w x Ž .6 A.K. Roy, B.M. Deb, Phys. Lett. A 234 1997 465.w x Ž .7 A.K. Roy, B.M. Deb, Chem. Phys. Lett. 242 1998 461.w x Ž .8 A. Nagy, I. Andrejkovics, J. Phys. B 27 1994 233.w x Ž .9 M.K. Harbola, V. Sahni, Phys. Rev. Lett. 62 1989 489.

w x Ž .10 V. Sahni, Y. Li, M.K. Harbola, Phys. Rev. A 45 19921434.

w x Ž .11 G. Brual, S.M. Rothstein, J. Chem. Phys. 69 1978 1177.w x Ž .12 C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 1988 785.w x13 J.C. Slater, Quantum Theory of Atomic Structure, Vol. II,

McGraw-Hill, New York, 1960.w x Ž .14 J.H. Wood, J. Phys. B 13 1980 1.w x15 T. Ziegler, A. Rauk, E.J. Baerends, Theor. Chim. Acta 43

Ž .1977 261.

w x Ž .16 R.M. Dickson, T. Ziegler, Int. J. Quant. Chem. 58 1996681.

w x17 A.C. Stuckl, C.A. Daul, H.U. Gudel, J. Chem. Phys. 107¨ ¨Ž .1997 4606.

w x18 C. Pollak, A. Rosa, E.J. Baerends, J. Am. Chem Soc. 119Ž .1997 7324.

w x19 M. Lannoo, G.A. Baraff, M. Schluter, Phys. Rev. B 24¨Ž .1981 943.

w x20 H. Tatewaki, T. Koga, Y. Sakai, A.J. Thakkar, J. Chem.Ž .Phys. 101 1994 4945.

w x21 E.R. Davidson, S.A. Hagstrom, S.J. Chakravorty, V.M. Umar,Ž .C.F. Fischer, Phys. Rev. A 44 1991 7071.

w x22 M.T. Carroll, R.F.W. Bader, S.H. Vosko, J. Phys. B 20Ž .1987 3599.

w x23 C.E. Moore, Atomic Energy Levels, Vol. I, National Bureauof Standards, Washington DC, 1949.

w x24 K.M. Dunseath, W.C. Fon, V.M. Burke, R.H.G. Reid, C.J.Ž .Noble, J. Phys. B 30 1997 277.

w x Ž .25 A.D. Becke, Phys. Rev. A 38 1988 3098.w x Ž .26 U. Kaldor, S.B. Ben-Schlomo, J. Chem. Phys. 92 1990

3680.w x Ž .27 J.F. Stanton, R.J. Bartlett, J. Chem. Phys. 98 1993 7029.w x Ž .28 H. Nakatsuji, Int. J. Quant. Chem. Symp. 17 1983 241.w x29 M. Petersilka, U.J. Gossmann, E.K.U. Gross, in: J.F. Dob-

Ž .son, G. Vignale, M.P. Das Eds. , Electronic Density Func-tional Theory: Recent Progress and New Directions, Plenum,New York, 1998.

w x30 C. Jamorski, M.E. Casida, D.R. Salahub, J. Chem. Phys. 104Ž .1996 5134.

w x31 M.E. Casida, C. Jamorski, K.C. Casida, D.R. Salahub, J.Ž .Chem. Phys. 108 1998 4439.

w x Ž .32 R. Bauernschmitt, R. Ahlrichs, Chem. Phys. Lett. 256 1996454.

w x Ž .33 S. Grimme, Chem. Phys. Lett. 259 1996 128.w x Ž .34 R. Singh, B.M. Deb, Phys. Rep. 1998 in press.