Can orbital-free density functional theory simulate molecules?

Junchao Xia, Chen Huang, Ilgyou Shin, and Emily A. Carter

Citation: J. Chem. Phys. 136, 084102 (2012); doi: 10.1063/1.3685604

View online: http://dx.doi.org/10.1063/1.3685604
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v136/i8
Published by the American Institute of Physics.

Additional information on J. Chem. Phys.

Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 THE JOURNAL OF CHEMICAL PHYSICS 136, 084102 (2012)

Can orbital-free density functional theory simulate molecules?

Junchao Xia,1 Chen Huang,2 Ilgyou Shin,3 and Emily A. Carter1,4,a)
1
Department of Mechanical and Aerospace Engineering, Princeton University,
Princeton, New Jersey 08544, USA
2
Department of Physics, Princeton University, Princeton, New Jersey 08544, USA
3
Department of Chemistry, Princeton University, Princeton, New Jersey 08544, USA
4
Program in Applied and Computational Mathematics Princeton University, Princeton, New Jersey 08544,
USA and the Gerhard R. Andlinger Center for Energy and the Environment, Princeton University, Princeton,
New Jersey 08544, USA
(Received 27 November 2011; accepted 26 January 2012; published online 22 February 2012)

Orbital-free density functional theory (OFDFT), with its attractive linearly scaling computation cost
and low prefactor, is one of the most powerful first principles methods for simulating large systems
(∼104 –106 atoms). However, approximating the electron kinetic energy with density functionals
limits the accuracy and generality of OFDFT compared to Kohn-Sham density functional theory
(KSDFT). In this work, we test whether the Huang-Carter (HC) kinetic energy density functional
(KEDF), which contains the physics to properly describe covalently bonded semiconductor materi-
als, can also be used to describe covalent bonds in molecules. In particular, we calculate a variety of
homonuclear diatomic molecules with the HC functional within OFDFT. The OFDFT bond dissoci-
ation energy, equilibrium bond length, and vibrational frequency of these dimers are in remarkably
good agreement with benchmark KSDFT results, given the lack of orbitals in the calculation. We
vary the two parameters λ (controlling the reduced density gradient contribution to the nonlocal ker-
nel) and β (the exponent of the density in the nonlocal term) present in the HC KEDF and find
that the optimal λ correlates with the magnitude of the highest occupied molecular orbital - lowest
unoccupied molecular orbital energy gap. Although the HC KEDF represents a significant improve-
ment over previous KEDFs in describing covalent systems, deficiencies still exist. Despite the sim-
ilar overall shape of the KSDFT and OFDFT ground state electron densities, the electron density
within the bonding region is still quite different. Furthermore, OFDFT is not yet able to give rea-
sonable description of magnetic states. The energy orderings of the triplet and singlet states of Si2
and Al family dimers are not consistent with KSDFT or experimental results and the spin polariza-
tion distributions also differ widely between the two theories. © 2012 American Institute of Physics.
[http://dx.doi.org/10.1063/1.3685604]

I. INTRODUCTION is proportional to the KSDFT computation cost in each sub-

domain. The prefactor will be even larger if k-point sampling
Among first principles quantum mechanics methods for
is needed when periodic boundary conditions are applied to
studying molecular and material properties, the most widely
sub-domains. As a result, in practice it is still prohibitive to
employed one today is density functional theory (DFT), the
carry out KSDFT simulations to answer interesting scientific
cornerstone of which was established by Hohenberg and
questions if the sample size is much beyond a few hundred
Kohn.1 One DFT scheme in particular, Kohn-Sham DFT
atoms, even with access to supercomputing power.
(KSDFT),2 has become one of the most powerful methods
An alternative DFT scheme, orbital-free DFT (OFDFT),9
available due to its ability to deliver quite accurate predic-
demonstrates promising competence in simulating much
tions at a reasonable computational cost. However, the orbital
larger numbers of atoms.10–13 Because the only variable in
orthonormalization and k-point sampling required in KSDFT
OFDFT is the density distribution of electrons, the number
makes the computation scale as the cube of the system size N,
of degrees of freedom is reduced from 3N (in the case of N
often with a large prefactor. Although a number of linear scal-
orbitals) to only 3 (for the electron density). This tremen-
ing KSDFT algorithms exist,3–8 they generally only become
dous simplification allows the computation to scale quasi-
linear scaling above ∼100 atoms due to a large algorithmic
linearly with the system size (O(NlnN)). In contrast to the
prefactor and are usually applicable only for nonmetallic sys-
linear scaling KSDFT methods, OFDFT has a much smaller
tems. Even though a couple of metallic/semi-metallic systems
prefactor,10 and shows excellent accuracy and efficiency when
have been studied with linear scaling KSDFT,8 a consider-
simulating metallic systems.10–15 OFDFT simulations of sam-
able buffer size is required to reach an accuracy comparable
ple sizes containing thousands and even 104 atoms are now
to conventional KSDFT methods. As a result, the prefactor
routine,11, 14, 15 with benchmark calculations performed for
of the linear scaling algorithm can become very large, which
more than 106 atoms, on a modest number of processors.10
a) Author to whom correspondence should be addressed. Electronic mail: As usual, however, a tradeoff exists between efficiency
eac@princeton.edu. and accuracy. Despite its attractive quasilinear scaling cost,

0021-9606/2012/136(8)/084102/13/$30.00 136, 084102-1 © 2012 American Institute of Physics

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-2 Xia et al.
OFDFT is thus far less accurate than and not as generally
applicable as KSDFT. The inaccuracy compared to KSDFT
originates from two sources. Unlike KSDFT in which orbital-
based nonlocal pseudopotentials (NLPSs) can be employed to
account for the interaction between electrons and ions (nuclei
plus their core electrons),16–18 the lack of orbitals in OFDFT
means the only option is to use a local pseudopotential (LPS),
which provides much less flexibility. However, some LPSs
have demonstrated excellent accuracy and transferability,19–21
even for a transition metal,22 when compared to NLPSs, sug-
gesting that inaccuracies in the electron-ion term in principle
can be overcome for OFDFT. Instead, the major source of er-
ror in OFDFT lies in describing the non-interacting electron
kinetic energy in terms of only the electron density, using a ki-
netic energy density functional (KEDF). Although the Hohen-
berg and Kohn theorems1 proved the existence of a universal
KEDF, they do not offer any specific details for constructing
it. The exact form of the KEDF remains unknown, although
many approximations to it have been proposed through years.
The Thomas-Fermi (TF) KEDF,23–25 which of course pre-
ceded modern DFT, was the first “naïve” attempt but it is only
exact for the non-interacting uniform electron gas and fails to
predict any atomic shell structure or chemical bonding.26, 27
Inclusion of the von Weizsäcker (vW) KEDF (Refs. 28–30)
(the TFλvW model) improves the TF model, but still is not
accurate for most systems. Other local or semilocal KEDF
models, containing higher order derivatives of the density,
were proposed later on but offered little improvement.31, 32
In recent decades, several nonlocal KEDFs, such as the
Chacón-Alvarellos-Tarazona,33–35 Wang-Teter (WT),36 and
Wang-Govind-Carter (WGC) (Ref. 37) KEDFs have been
proposed, which are all based on linear response theory. These
KEDFs exhibit greatly improved accuracy compared to lo-
cal or semilocal KEDFs, and demonstrated accuracy com-
parable to KSDFT for nearly-free-electron-like main group
metals.38–40 However, constructing an accurate KEDF for sys-
tems other than main group metals remained elusive. For ex-
ample, the aforementioned nonlocal KEDFs are mostly in-
adequate for studying covalently bonded materials,41 where
valence electrons are more localized and the linear response
behavior is rather different from that in metallic systems.42
Very recently, the Huang-Carter (HC) KEDF was pro-
posed for semiconductor materials based on the dielectric
response of semiconductors. It exhibited remarkable accu-
racy and transferability in calculating properties of silicon and
group III/V semiconductors.43 This motivated us to consider
if this HC KEDF could improve the generality of OFDFT
such that other types of systems could be accurately treated. In
particular, because the HC KEDF is able to treat the covalent
bonds in crystals, we sought to test the validity of OFDFT
in treating covalent bonds in molecules. Here we focus on
homonuclear diatomics (dimers) so as to study purely cova-
lent bonds, though one example of a heteronuclear diatomic
is also provided.
Despite their seeming simplicity, quantitative treatment
of dimers remains a huge challenge for OFDFT. To our
knowledge, very few studies44 have been done on molecule
dissociation using self-consistent OFDFT calculations. Some
have employed Hartree-Fock or KSDFT densities to evaluate
Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
J. Chem. Phys. 136, 084102 (2012)
non-self-consistent OFDFT bond energies in small molecules
such as F2 and the CH4 molecule.45–47 All of them, how-
ever, demonstrated unsatisfactory OFDFT results due to in-
accurate KEDFs such as those based on a gradient expansion.
On the other hand, the nonlocal KEDFs proposed for nearly-
free-electron-like metals based on the perturbed uniform elec-
tron gas are also not appropriate for describing molecules.
A simple dimer is even more difficult for OFDFT to treat
than semiconductor crystals, because multiple covalent bonds
can be involved and spin-polarized calculations are also re-
quired for the open-shell ground states of some dimers. In
the present work, a variety of dimers are calculated utilizing
OFDFT with the HC KEDF. Via comparison with KSDFT re-
sults, we demonstrate that OFDFT can treat dimer molecules
remarkably well given the lack of a wavefunction, although
OFDFT clearly still has remaining defects to be rectified.
The rest of this paper is structured as follows. In
Secs. II and III, the formalisms and numerical details are
given. Then results for properties of different kinds of dimers
are presented, including equilibrium bond lengths, bond dis-
sociation energies, and vibrational frequencies. Predictions
made with the HC and other semilocal and nonlocal KEDFs
are compared to KSDFT benchmarks. Ground state density
distributions are also analyzed and compared with KSDFT
densities. The sensitivity of the results to the parameters in
the HC KEDF is also presented and discussed. We end with
a discussion of prospects for future improvements.
II. FORMALISM
The Hohenberg-Kohn theorems1 state that the electronic
total energy can be expressed as a functional of the electron
density alone:
E[n] = Ts [n] + J [n] + Exc [n] + Eext [n], (1)
where n is the total electron density, Ts is the non-interacting
electron kinetic energy, J is the Hartree electron repulsion
energy, and Exc is the exchange-correlation energy. The last
term, Eext , is the energy related to external fields, such as ionic
(or nuclear) potentials. This Hohenberg-Kohn energy func-
tional in Eq. (1) is precisely the OFDFT energy functional.
A number of the dimers have ground states that are open-
shell multiplets. To treat them, we must extend the OFDFT
formalism to allow for spin polarization. The total OFDFT
energy functional including spin polarization is simply writ-
ten as
E[nup , ndown ] = Ts [nup , ndown ] + J [nup , ndown ]
+Exc [nup , ndown ] + Eext [nup , ndown ],
(2)
where nup and ndown are the densities of spin up and spin
down electrons, respectively. The Hartree and external en-
ergies generally remain the same as in the spin-unpolarized
formalism, though local spin-dependent electron-ion poten-
tials could be used.48 The exchange-correlation energy can
be evaluated within local spin density approximation (LSDA)
(Refs. 49–52) or spin-polarized generalized gradient approx-
imation (GGA) functionals.53 The non-interacting kinetic
084102-3 Orbital-free DFT for molecules
energy is evaluated through the equation,54
1 1
Ts [nup , ndown ] = Ts [2nup ] + Ts [2ndown ]. (3)
2 2
The total energy is variationally minimized subject to the con-
straint that the total number of electrons remains fixed. The
problem of finding a minimum with a constraint can be con-
verted to a problem of finding a stationary point by means of
Lagrange multipliers:
  
δ E[n] − μ n(r)dr − N = 0, (4)
where N is the total electron number and μ is the chemical
potential. For spin-polarized and fixed magnetization calcula-
tions, we simply generalize to
 
δ E[nup , ndown ] − μup nup (r)dr − Nup

− μdown ndown (r)dr − Ndown
where Nup and Ndown are the spin up and spin down electron
numbers, and μup and μdown are the chemical potentials for
spin up and spin down electrons, respectively. Our optimiza-
tion algorithm is implemented similarly to that described in
Ref. 55, except that in each iteration the optimization direc-
tions θ α and θ β are simultaneously optimized to minimize
the total energy via a two-dimensional conjugate gradient
search instead of the approximation introduced in the origi-
nal paper.55
III. NUMERICAL DETAILS
The OFDFT calculations are performed with a modified
version of our PROFESS 2.0 code56, 57 and the benchmark KS-
DFT calculations are carried out with the ABINIT code.58
In both OFDFT and KSDFT calculations, bulk-derived lo-
cal pseudopotentials (BLPSs) are used. For all local density
approximation (LDA) BLPSs except that of lithium, we em-
ploy previously reported ones,21, 43 while the lithium LDA
BLPS and all GGA BLPSs were generated as described in
the literature;21 details are given in the Appendix. For the
exchange-correlation functionals, the Perdew-Zunger (PZ)
(Refs. 49, 50, and 52) form of the LSDA and the Perdew-
Burke-Ernzerhof (PBE) (Ref. 53) form of the GGA is em-
ployed in all calculations.
As mentioned earlier, our focus is on testing the HC
KEDF for use in molecular simulations. The HC KEDF pos-
sesses a similar form to some previous nonlocal KEDFs:36, 37
Ts [n] = TTF [n] + TvW [n] + TNL [n],
 (ωe ) and bond dissociation energy (D0 ) is computed and then
TTF [n] = CTF n(r)5/3 dr,

1 ∇n(r) · ∇n(r)
TvW [n] = dr,
8 n(r)
Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
J. Chem. Phys. 136, 084102 (2012)

TNL [n] = C n(r)8/3−β ω(r, r )n(r )β drdr , (9)
where TTF is the TF KEDF with CTF
= 3/10(3π 2 )2/3 ; TvW is the vW KEDF, and TNL is the
nonlocal term, which includes a single-density-dependent
kernel:
ω(r, r ) = ω[ξ (r, r )|r − r |], (10)
ξ (r, r ) = kF (r)(1 + λs(r)2 ), (11)
|∇n(r)|
s(r) = , (12)
n(r)4/3
5
−βηω̃(η) + (5 − 3β)β ω̃(η) = [F (η) − 3η2 − 1], (13)
3
= 0, (5) where kF is the Fermi wave vector, kF (r) = (3π 2 n(r))1/3 .
The kernel can be solved for numerically in reciprocal space
according to Eq. (13), where η is a dimensionless momen-
tum vector, η = q/(2kF ). For the parameters λ and β in the
HC KEDF, selected values in the literature43 (λ = 0, 0.01,
and 0.01177, β = 0.65 and 0.7143) are employed. For some
dimers, the parameters are tuned slightly to achieve better
results. We also compare the accuracy of the HC KEDF to
the TF1/5vW,29, 59 TF1/9vW,31 WGC,37 and WT (Ref. 36)
KEDFs. When using WGC and WT KEDFs, an average elec-
tron density needs to be specified, which is not well defined
for molecules. Here we first obtain the ground state density
from KSDFT calculations at the equilibrium bond length of
each dimer. The average density is then determined by av-
eraging the total electron density in the bonding region be-
tween nuclei, where the density is greater than 10−2 a.u. The
kinetic energy cutoff for the plane-wave basis is selected so
that the total energy is converged to within 1 meV/atom.
KSDFT calculations all employ a 900 eV kinetic energy cut-
off and OFDFT calculations use a kinetic energy cutoff of
1600 eV. In each total energy vs. bond length curve, the ini-
tial guess for the wavefunctions in KSDFT calculations is set
to be the default in the ABINIT code for the first bond length,
usually 1.8 Å. Upon increase of the bond length, for each of
the subsequent bond length calculations, the initial guess is
taken as the resulting wavefunctions from the previous (one
step smaller) bond length calculation. In OFDFT calculations,
the initial guess of the density is always taken as the uniform
average density of the system (defined as discussed above).
To test the quality of OFDFT for treating diatomic
molecules, different kinds of dimers including Al2 , Ga2 , In2 ,
Si2 , P2 , As2 , Sb2 , Li2 , and Mg2 are considered. A heteronu-
(6) clear diatomic molecule, AlP, is also examined. For each of
them, the equilibrium bond length (re ), vibrational frequency
(7) compared with KSDFT results. In both OFDFT and KSDFT
calculations, the periodic cell is set to be 20 × 10 × 10 Å, with
the dimer in the center of the cell aligned in the longest direc-
tion, which guarantees the distance between nearest images is
(8) larger than 10 Å for all bond lengths considered. In generating
084102-4 Xia et al.
the total energy versus bond length curves, the dimer is kept
in the center while the distance between two atoms is varied.
The magnetization is fixed to be 0 in the singlet calculations
and 2 in the triplet calculations. Around the total energy mini-
mum, a ±0.01 Å region is employed to perform quadratic fit-
ting so as to calculate the equilibrium bond length re and the
vibrational frequency ωe , which is then used to calculate the
zero-point energy, E0 = hωe /4π . To calculate the bond dis-
sociation energy, the difference between the total energies at
the equilibrium bond length and the dissociation limit (bond
length equal to 10 Å) is first computed, and then we subtract
the zero-point energy to obtain the D0 values reported. Single
atom energies are also calculated, with a 10 × 10 × 10 Å pe-
riodic cell. The resultant energy is then multiplied by two to
obtain the full dissociation limit energy.
IV. RESULTS AND DISCUSSION
A. Energy analysis
Table I lists results for Al2 , Ga2 , and In2 , for which the
triplet state is the ground state for all three dimers; the ex-
act ground state electron configuration has been under debate
J. Chem. Phys. 136, 084102 (2012)
down electron occupies each of the π -orbitals, respectively,
resulting in an open-shell singlet. However, due to the single
determinant nature of KSDFT, it actually cannot be consid-
ered as a true 1 g state.
The parameters in the HC KEDF in Table I were ad-
justed to generate reasonable D0 values, while re changes lit-
tle with these parameter alterations. For these three dimers,
the HC KEDF parameters can be adjusted such that OFDFT
D0 results are in remarkably good agreement with the KS-
DFT benchmarks. In all cases, λ = 0 gave the best results,
while β = 0.51 or 0.7143 were optimal. For singlet states, the
D0 difference between OFDFT and KSDFT is usually less
than 0.03 eV, while for triplet states, the difference is a little
larger, about 0.3 eV. The re values computed by OFDFT are
very close to the KSDFT values, with deviations usually less
than 0.05 Å (except for the GGA results of In2 ). The OFDFT
vibrational frequencies are also in a reasonable range around
the KSDFT values. We observe the usual overbinding by LDA
in both KSDFT and OFDFT results: LDA generally predicts
larger D0 , smaller re , and larger ωe than GGA does. Remark-
ably, OFDFT-GGA predicts D0 , re , and ωe within 0.15 eV,
0.05 Å, and 30 cm−1 of experiment, respectively.
To put this level of agreement in perspective, Table II
for a long time, with two candidates 3u and 3
g − nearly compares the ability of different KEDFs used within OFDFT
degenerate.60–72 Some recent experiments and calculations fa- to predict D0 . TF KEDF results are not listed because (as ex-
vor 3 u as the ground state.62–65, 69, 70 Here we obtained a
3
g − ground state for Al2 and Ga2 with both LSDA and spin-
polarized GGA exchange-correlation functionals, consistent
with some previous KS-DFT calculations.66, 68 For In2 , LSDA
pected) it did not produce bonding for any dimer. Also, WGC
KEDF results could not be obtained because the evaluation of
the WGC KEDF via Taylor expansion for highly fluctuating
densities makes the calculation unstable and hard to converge.
calculations predict a 3 u ground state while spin-polarized Although all the other KEDFs examined predict bound states,
GGA calculations give a 3
g − ground state. The table also the TFλvW KEDFs greatly underbind the molecules while the
includes singlet state results for comparison. In the electron WT KEDF greatly overbinds them. The HC KEDF is the only
configuration of the singlet state, one spin up and one spin one that can obtain near quantitative agreement with KSDFT
for covalent bond energies, albeit without a universal param-
TABLE I. OFDFT, KSDFT, and experimental bond dissociation energies eter set (optimal λ or β varies somewhat).
(D0 ), equilibrium bond lengths (re ), and vibrational frequencies (ωe ) for Al2 , Despite this achievement, Fig. 1 reveals the first clue of
Ga2 , and In2 in singlet (Ms = 0) and triplet (Ms = 1) states. λ and β are remaining flaws in our formalism. The KSDFT and OFDFT
the parameters in the HC KEDF (see text). The optimal λ = 0 for all three potential energy curves are compared in Fig. 1 for the two
molecules. OFDFT values are listed first while KSDFT values are listed in
parentheses.
different spin states of Al2 . KSDFT (black and red squares)
predicts the triplet state to be lower in energy than the sin-
D0 re ωe glet state, consistent with experiment. OFDFT, independent
Dimer β (eV) (Å) (cm−1 ) of KEDF used, predicts the reverse energy ordering, with the
Expt. Al2 (Ms = 1) ... 1.55a 2.466a 350a
Ga2 (Ms = 1) ... 1.40a 180b
In2 (Ms = 1) ... 1.01a 118b TABLE II. D0 values of Al2 , Ga2 , In2 , and Si2 in singlet (Ms = 0) and
triplet (Ms = 1) states calculated by KSDFT, OFDFT with HC, TFλvW
LDA Al2 (Ms = 0) 0.51 1.72 (1.74) 2.498 (2.473) 285 (346) (λ = 1/5 and 1/9) and WT KEDFs. All values are in eV. LDA exchange-
Al2 (Ms = 1) 0.7143 1.70 (2.00) 2.489 (2.464) 324 (351) correlation is used.
Ga2 (Ms = 0) 0.51 1.67 (1.69) 2.329 (2.323) 208 (212)
Ga2 (Ms = 1) 0.7143 1.64 (1.96) 2.312 (2.312) 211 (216) OFDFT OFDFT OFDFT OFDFT
In2 (Ms = 0) 0.51 1.64 (1.64) 2.669 (2.644) 139 (154) Dimer KSDFT (HC) (TF1/5vW) (TF1/9vW) (WT)
In2 (Ms = 1) 0.7143 1.65 (1.87) 2.661 (2.633) 148 (157)
Al2 (Ms = 0) 1.74 1.72 0.73 0.36 7.99
GGA Al2 (Ms = 0) 0.51 1.50 (1.49) 2.543 (2.497) 309 (328)
Al2 (Ms = 1) 2.00 1.70 0.64 0.29 13.44
Al2 (Ms = 1) 0.7143 1.49 (1.80) 2.519 (2.490) 321 (343)
Ga2 (Ms = 0) 1.69 1.67 0.63 0.27 10.24
Ga2 (Ms = 0) 0.51 1.29 (1.28) 2.447 (2.422) 165 (192)
Ga2 (Ms = 1) 1.96 1.64 0.52 0.21 16.73
Ga2 (Ms = 1) 0.7143 1.25 (1.60) 2.431 (2.411) 175 (193)
In2 (Ms = 0) 1.64 1.64 0.63 0.29 6.66
In2 (Ms = 0) 0.7143 1.15 (1.18) 2.775 (3.131) 102 (89)
In2 (Ms = 1) 1.87 1.65 0.56 0.24 12.60
In2 (Ms = 1) 0.7143 1.05 (1.31) 2.798 (3.080) 100 (94)
Si2 (Ms = 0) 4.59 4.86 0.84 0.33 13.82
a
Reference 71. Si2 (Ms = 1) 4.66 4.66 0.73 0.26 28.11
b
Reference 72.

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-5 Orbital-free DFT for molecules J. Chem. Phys. 136, 084102 (2012)

FIG. 1. Total energy vs. bond length (r) curves for singlet and triplet Al2 FIG. 2. Total energy vs. bond length curves for singlet and triplet Si2 from
from KSDFT and OFDFT with the TF1/9vW KEDF and the HC KEDF (λ KSDFT and OFDFT with TF1/9vW KEDF and HC KEDF (λ = 0.01 and β
= 0 and β = 0.65). All calculations use the LDA exchange-correlation func- = 0.65). All calculations use the LDA exchange correlation functional.
tional. The KSDFT curves exhibit a slight jump just beyond re when the en-
ergy ordering of the σ - and π -MOs flips, while the OFDFT curves are smooth
everywhere.
as face-centered-cubic (fcc) bulk Al, while non-zero λ is
optimal for covalent systems, such as bulk cubic diamond
triplet state much higher in energy than the singlet state. Here (CD) Si. As it turns out, the optimal parameter set for Si2 ,
we used for both spin states a value of β intermediate between λ = 0.01 and β = 0.65, is the same as that for bulk CD
the two optimal values in the HC KEDF. Adjusting β over a Si.43 The coincidence is interesting although these dimers
large range (0.51–1.0) did not change the ordering of the two are molecular rather than bulk crystals. We will return to the
spin states. Other KEDFs such as TF1/9vW produces similar effects of parameters λ and β in Sec. IV C.
energy orderings, as shown in Fig. 1. The origin of this inac- Table IV shows predicted properties of P2 , As2, and Sb2 ,
curacy will be discussed further when analyzing the electron which all form triply bonded singlet (1 g + ) ground states.
density distribution in Sec. IV B. For dimers of this family, although re does not change much
The situation is quite similar for Si2 . The ground state is when the HC KEDF parameters are adjusted, zero or small λ
also a triplet (3 g − ) (Ref. 71) state. Table III displays the pre- yields too small Do and ωe values compared to KSDFT. The
dicted properties for Si2 and Fig. 2 provides the correspond- OFDFT values increase with λ, however even with λ equal
ing potential energy curves. Again, the OFDFT HC KEDF to 0.03, OFDFT still underbinds these dimers, producing
and KSDFT D0 , re , and ωe values match reasonably well but smaller D0 , larger re , and smaller ωe than KSDFT. Further
OFDFT still incorrectly predicts the ground state to be sin- increases of λ offer little improvement, with OFDFT D0
glet instead of triplet. Nevertheless, Table II reveals that the values ∼2 eV smaller than KSDFT values. OFDFT re values
HC KEDF is the best among the KEDFs, with TFλvW again are ∼0.2–0.3 Å larger than those of KSDFT and the vibra-
greatly underbinding and WT vastly overbinding. tional frequencies are smaller in OFDFT than in KSDFT by
More attention could be paid to the choice of parameters ∼100–200 cm−1 . Despite the lack of quantitative agreement
in the HC KEDF for the Al family dimers and Si2 . We
see that zero λ is optimal for the Al family dimers while
non-zero λ is optimal for Si dimer. In the HC KEDF paper,43 TABLE IV. OFDFT, KSDFT, and experimental D0 , re , and ωe for P2 , As2 ,
zero λ was found to be optimal for metallic systems, such and Sb2 . OFDFT values are listed first while KSDFT values are listed in
parentheses. λ = 0.03, β = 0.51 is used in the HC KEDF.
TABLE III. OFDFT, KSDFT, and experimental D0 , re , and ωe for Si2 in
singlet (Ms = 0) and triplet (Ms = 1) states. OFDFT values are listed first D0 re ωe
while KSDFT values are listed in parentheses. λ = 0.01, β = 0.65 is used in Dimer (eV) (Å) (cm−1 )
the HC KEDF.
Expt. P2 5.03a 1.89a 781a
D0 re ωe As2 3.96a 2.10a 430a
Dimer (eV) (Å) (cm−1 ) Sb2 3.09a 2.34a 270a
LDA P2 7.12 (9.54) 2.138 (1.942) 628 (790)
Expt. Si2 (Ms = 1) 3.21a 2.246a 510.98a As2 6.12 (8.35) 2.247 (2.032) 352 (459)
LDA Si2 (Ms = 0) 4.86 (4.59) 2.246 (2.284) 513 (501) Sb2 4.71 (6.65) 2.710 (2.431) 212 (283)
Si2 (Ms = 1) 4.66 (4.66) 2.259 (2.277) 486 (509) GGA P2 6.74 (9.38) 2.124 (1.915) 622 (812)
GGA Si2 (Ms = 0) 4.48 (3.93) 2.283 (2.309) 469 (489) As2 5.49 (7.62) 2.324 (2.084) 327 (425)
Si2 (Ms = 1) 4.30 (4.15) 2.296 (2.303) 487 (492) Sb2 4.57 (5.92) 2.706 (2.459) 213 (261)
a a
Reference 71. Reference 71.

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-6 Xia et al. J. Chem. Phys. 136, 084102 (2012)

TABLE V. D0 values of P2 , As2 , Sb2 , and Li2 calculated by KSDFT,

OFDFT with HC, TFλvW (λ = 1/5 and 1/9) and WT KEDFs. All values
are in eV. LDA exchange-correlation is used.

OFDFT OFDFT OFDFT OFDFT

Dimer KSDFT (HC) (TF1/5vW) (TF1/9vW) (WT)

P2 9.54 7.12 0.94 0.39 15.80

As2 8.35 6.12 0.66 0.20 12.91
Sb2 6.65 4.71 0.71 0.31 4.73
Li2 1.45 0.97 0.33 0.24 1.61

with KSDFT, the covalent bonding physics contained in the

HC KEDF yields far superior results to those provided by
TFλvW and WT KEDFs, as seen in Table V. However, the
triple bond character is clearly more challenging to capture in
OFDFT than the single or double bonds of group III and IV
dimers.
Finally, we consider Li2 and Mg2 with results listed in Ta-
ble VI. OFDFT fortuitously predicts a D0 in excellent agree-
ment with experiment for Li2 , but it is ∼30% smaller than
KSDFT, which it is meant to approximate. The bond lengths
are overestimated while the vibrational frequencies are under-
estimated for Li2 within OFDFT, with respect to both KSDFT
and experiment. For Mg2 , OFDFT predicts a much larger D0
than KSDFT and experiment, which is actually not surpris-
ing. Since each of the Mg atoms has two 3s electrons and
forms a closed-shell structure, the Mg dimer only forms a van
der Waals bond. The poor prediction of OFDFT for Mg2 is
expected because the HC KEDF is not designed to describe FIG. 3. The general trend of D0 and re values from OFDFT and KSDFT.
the van der Waals interaction. Furthermore, it is known that (a) singlet state (Ms = 0); (b) triplet state (Ms = 1), all calculations use the
PBE GGA exchange-correlation functional.
KSDFT with LDA or GGA does not treat the van der Waals
interaction accurately, which is reflected in the significant dif-
ference between the KSDFT and experimental values. states, are nearly degenerate in energy. The single determi-
The trends in PBE GGA D0 and re for all dimers except nant description in KSDFT cannot treat this kind of multi-
Mg2 are summarized in Fig. 3. Mg2 is not included since reference case properly. This can be seen in the large differ-
the van der Waals interaction is not our focus here. We ob- ence between KSDFT and experimental D0 values, especially
serve that OFDFT generally produces similar trends for D0 for the group V family dimers. Instead, single atom calcula-
as KSDFT. The D0 values for Al, Ga, In, and Si dimers are tions can be carried out and their total energies (multiplied
reasonably close to the KSDFT ones, independent of spin by two) could be considered as the total energy at the fully
state, while OFDFT predicts smaller D0 for Li, P, As, and Sb dissociated limit. The corrected D0 values computed in this
dimers. For re , OFDFT values fluctuate around the KSDFT way are listed in Table VII. The results from OFDFT calcu-
values, but the deviation is not large, at most ∼0.3 Å off the lated in the same way are also listed. We see that for Li2 and
KSDFT values. the group III family dimers, OFDFT results are still in rather
We note that KSDFT has difficulty describing the dissoci- good agreement with the KSDFT and experimental values,
ation limit, where the molecular orbitals formed by p atomic while OFDFT predicts much larger D0 values for Si2 and the

TABLE VI. OFDFT, KSDFT, and experimental D0 , re , and ωe for Li2 and Mg2 . OFDFT values are listed first while
KSDFT values are listed in parentheses.

D0 re ωe
Dimer Parameters (eV) (Å) (cm−1 )

Li2 Expt. ... 1.05a 2.67a 351a

LDA λ = 0.01, β = 0.7143 0.97 (1.45) 3.125 (2.781) 237 (314)
GGA λ = 0.01, β = 0.7143 0.93 (1.42) 3.103 (2.775) 267 (314)
Mg2 Expt. ... 0.05a 3.89a 51a
LDA λ = 0, β = 0.65 1.43 (0.21) 2.740 (3.405) 224 (114)
GGA λ = 0, β = 0.65 1.28 (0.15) 2.807 (3.493) 223 (110)
a
Reference 71.

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-7 Orbital-free DFT for molecules J. Chem. Phys. 136, 084102 (2012)

TABLE VII. Experimental, OFDFT, and KSDFT D0 values calculated using the two single atom dissociation limit;
see text for details. All units are in eV. The HC KEDF with optimal λ and β values is used in OFDFT calculations.

Dimer OFDFT_LDA OFDFT_GGA KSDFT_LDA KSDFT_GGA Expt.

Al2 (Ms = 0) 2.61 2.30 1.76 1.50 ...

Al2 (Ms = 1) 1.70 1.49 1.96 1.66 1.55a
Ga2 (Ms = 0) 2.69 1.81 1.71 1.29 ...
Ga2 (Ms = 1) 1.64 1.27 1.92 1.45 1.4a
In2 (Ms = 0) 2.38 1.82 1.66 1.20 ...
In2 (Ms = 1) 1.65 1.05 1.84 1.23 1.01a
Si2 (Ms = 0) 8.34 7.68 3.87 3.43 ...
Si2 (Ms = 1) 7.30 6.72 4.05 3.61 3.21a
P2 15.89 15.29 6.31 5.57 5.03a
As2 14.34 13.10 5.23 4.02 3.96a
Sb2 10.92 10.14 4.02 2.83 3.09a
Li2 1.10 0.92 0.95 0.71 1.05a
Mg2 1.43 1.28 0.21 0.14 0.05a
a
Reference 71.

group V family dimers. This is because the HC KEDF is still to the KSDFT values. The HC KEDF results again are far su-
not accurate enough to describe open-shell multiplets, as al- perior to the other KEDFs.
ready demonstrated by the incorrect energy ordering between
singlet and triplet states. The total energy computed for a sin-
gle open-shell multiplet atom is inaccurate, leading to poor B. Electron density analysis
D0 values, especially for Si atom (Ms = 1) and the group V
To further evaluate the quality of OFDFT in describing
family atoms (Ms = 3/2). It is well known that single atom
molecules, it is also important to compare the self-consistent
calculations pose a real challenge for OFDFT calculations.73
ground state electron density between OFDFT and KSDFT
For now, we do not focus on the accuracy of OFDFT in calcu-
results. Figure 4 shows the total electron density along the
lating single open-shell atoms. Instead, OFDFT D0 values are
bond axis for Si2 at its equilibrium bond length in the sin-
calculated only from the potential energy curves at the dimer
glet state. KSDFT and OFDFT generate density distributions
dissociation limit (10 Å) and then compared to the consistent
with similar general shapes, but in the bonding region, the
KSDFT values.
KSDFT density is larger than what OFDFT predicts, quite
Besides homonuclear diatomic molecules, one heteronu-
similar to the situation in bulk silicon, as shown in Fig. 4 of
clear diatomic molecule, AlP, is also calculated and results
Ref. 43. The electron density comparisons for singlet Al2
are shown in Table VIII. It has been found in experiments and
and P2 are presented in Figs. 5 and 6, respectively. For Al2 ,
computations that two triplet states, 3 − and 3  are nearly
degenerate with 3 − slightly lower than 3 .74, 75 Here we
only report the triplet ground state, which is found to be a
3
 state. For the singlet state, similar to the group III dimers,
KSDFT tries to describe an open-shell singlet state, but it
is extremely difficult to converge because the required two-
determinant description is missing in the KSDFT formalism.
Through slight tuning of the parameters in the HC KEDF, rea-
sonably good D0 , re , and ωe could also be obtained compared

TABLE VIII. OFDFT, KSDFT, and experimental D0 , re , and ωe for AlP

(Ms = 1). λ = 0.003, β = 0.65 is used in the HC KEDF OFDFT calculations.

D0 re ωe
Method (eV) (Å) (cm−1 )

Expt. ... 2.20a ... ...

LDA KSDFT 2.92 2.233 539
OFDFT_HC 3.02 2.341 454
OFDFT_TF1/5vW 0.75 2.642 232
OFDFT_TF1/9vW 0.32 2.833 160
GGA KSDFT 2.57 2.235 530
OFDFT_HC 2.73 2.366 410 FIG. 4. Valence electron density of KSDFT and OFDFT along the bond axis
OFDFT_TF1/5vW 0.51 2.724 126 direction for singlet Si2 at its equilibrium bond length, 2.284 Å for KSDFT
OFDFT_TF1/9vW 0.23 3.294 148 (for KSDFT, the two Si nuclei are located at 8.858 Å and 11.142 Å) and 2.246
Å for OFDFT (for OFDFT, the two Si nuclei are at 8.877 Å and 11.123 Å).
a
Reference 71. The LDA exchange-correlation functional is used.

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-8 Xia et al. J. Chem. Phys. 136, 084102 (2012)

FIG. 5. Valence electron density of KSDFT and OFDFT along the bond axis
direction for singlet Al2 at its equilibrium bond length, 2.473 Å for KSDFT
(the two Al nuclei are at 8.764 Å and 11.237 Å) and 2.498 Å for OFDFT (the
two Al nuclei are at 8.751 Å and 11.249 Å). The LDA exchange-correlation
functional is used.

OFDFT has larger density than KSDFT in the bonding re-

gion. The OFDFT density in the bonding region is increased
here because a smaller λ (λ = 0) is used in the Al dimer calcu-
lations. Generally, a larger λ lowers the OFDFT density in the
bonding region, as will be discussed in Sec. IV C. This effect
is further illustrated in the electron density comparison for P2 ,
where KSDFT predicts a much larger density than OFDFT in
the bonding region.
A similar density comparison can also be carried out for
the triplet states of the dimers. Figure 7 demonstrates the spin
up and spin down electron density for triplet Si2. KSDFT
has larger density in the bonding region for both spin up

FIG. 7. (a) spin up and (b) spin down valence electron densities of KSDFT
and OFDFT along the bond axis direction for triplet Si2 at its equilibrium
bond length, 2.277 Å for KSDFT (the two Si nuclei are at 8.862 Å and 11.139
Å) and 2.259 Å for OFDFT (the two Si nuclei are at 8.871 Å and 11.130 Å).
The LDA exchange-correlation functional is used.

and spin down electrons. Unlike KSDFT, OFDFT does not

FIG. 6. Valence electron density of KSDFT and OFDFT along the bond axis
direction for singlet P2 at its equilibrium bond length, 1.942 Å for KSDFT
(the two P nuclei are at 9.029 Å and 10.971 Å) and 2.138 Å for OFDFT (the
two P nuclei are at 8.931 Å and 11.069 Å). The LDA exchange-correlation
functional is used.
have equivalent densities in both spin channels. Although
the general shapes look similar in KSDFT and OFDFT, the
discrepancy between the two can be seen most clearly by
comparing another important dimensionless variable, the
spin polarization ζ , which is calculated as
nup (r) − ndown (r)
ζ (r) = . (14)
nup (r) + ndown (r)
In Fig. 8, isosurfaces of different ζ values for triplet Si2 are
displayed such that we may compare how the isosurfaces
evolve with decreasing ζ . In KSDFT, when ζ is large, the
isosurface has a donut-like shape around the bond axis. This
makes complete sense because the extra two spin up electrons
comprising the triplet occupy two degenerate, orthogonal
π -orbitals. As the ζ value decreases, the isosurface expands.
However, in OFDFT, the shape of the isosurface is entirely

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-9 Orbital-free DFT for molecules J. Chem. Phys. 136, 084102 (2012)

FIG. 8. Spin polarization ζ iso-surfaces for triplet Si2 : KSDFT results for
(a) ζ = 0.7, (b) ζ = 0.4, and (c) ζ = 0.15. OFDFT results for (d) ζ = 0.7,
(e) ζ = 0.4, and (f) ζ = 0.15. All calculations employ the LDA exchange-
correlation functional.

different. Furthermore, the evolving trend is also reversed.

The isosurface is large when ζ is large and shrinks when ζ
reduced. The inaccurate behavior of OFDFT for triplet state
calculations is reflected not only through the incorrect energy
ordering between singlet and triplet dimers, but also through
the totally different spin polarization distributions from the
KSDFT results. This is likely the origin of the incorrect order
FIG. 9. Total energy vs. bond length curves of KSDFT and OFDFT with
of spin states and clearly requires further advances in the different HC KEDF parameter values. (a) Si2 in its singlet state; (b) Si2
LPSs and KEDFs to remedy. in its triplet state; and (c) P2 . All calculations employ the LDA exchange-
correlation functional.

C. Effects of the parameters

By tuning the HC KEDF parameters λ and β, the abso-
lute energy curve may shift upward or downward, as shown in
Fig. 9. Generally, increasing λ moves the whole curve upward
while increasing β moves the curve downward. This behavior
is consistent for all of the dimers, as demonstrated in Fig.
10, where the absolute total energies for all dimers at their
equilibrium bond lengths are plotted. For every parameter set
shown, the curve is shifted such that the OFDFT energy of
Ga2 matches the KSDFT value for Ga2 . The shifted curves
almost overlap with each other, which implies that despite
different absolute energy values with different parameter sets,
the general trend of the absolute energies of each dimer at its
equilibrium bond length is similar, both for singlet and triplet
calculations.
The behavior displayed in Figs. 9 and 10 indicates that
there could be a universal parameter set that could produce
a total energy at re close to the KSDFT values for every
dimer. However, as mentioned earlier, the parameter choice
(especially λ) greatly affects the D0 and ωe values, as well
as the ground state electron density distribution. Table IX
shows results for Si2 using different parameter values in the
HC KEDF with the LDA exchange-correlation functional. Al-
though changing parameters has only a small effect on re (dif-
fering less than 0.1 Å), D0 and ωe are much more sensitive to
the parameter choice, particularly between zero and non-zero
λ parameter sets. Non-zero λ tends to produce a deep well,
leading to large D0 and ωe , while zero λ predicts a shallow

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-10 Xia et al. J. Chem. Phys. 136, 084102 (2012)

FIG. 10. Absolute total energy of each dimer at its equilibrium bond length.
OFDFT curves are shifted to guarantee that energy values of Ga2 matches
with the KSDFT results. (a) singlet state dimers; (b) triplet state dimers. All
calculations employ the LDA exchange-correlation functional.
FIG. 11. (a) spin up and (b) spin down electron density for triplet Al2 at its
equilibrium bond length with different λ values in the HC KEDF. β is fixed
to be 0.7143. The LDA exchange-correlation functional is used.
TABLE IX. D0 , re , and ωe results of KSDFT and OFDFT with different
parameter values for Si2 . The LDA exchange-correlation functional is used.

D0 re ωe well, which can also be seen in Fig. 9. Generally, the larger

Method (eV) (Å) (cm−1 ) λ is, the larger D0 and ωe will be. Increasing the other pa-
rameter, β, also slightly raises the D0 values, but much less
Singlet
KS 4.59 2.284 501
dramatically.
OFDFT Besides D0 , re , and ωe values, we also analyze the effect
λ = 0, β = 0.65 1.80 2.307 377 of changing the HC KEDF parameters on the electron density
λ = 0, β = 0.7143 1.78 2.296 395 distributions, as shown in Fig. 11 for triplet Al2 with differ-
λ = 0.01, β = 0.65 4.86 2.246 513 ent λ values. As mentioned in Sec. IV B, the reason why the
λ = 0.01, β = 0.7143 4.94 2.235 521 density comparison between OFDFT and KSDFT has contra-
λ = 0.01177, β = 0.65 5.01 2.247 518 dictory behavior for Si2 versus Al2 is the different parameter
λ = 0.01177, β = 0.7143 5.10 2.236 527 sets employed for these two dimers. For Al2 , the OFDFT den-
Triplet sity with λ equal to zero is larger than the KSDFT density in
KS 4.66 2.277 509
the bonding region. If λ is increased, the OFDFT density in
OFDFT
the bonding region decreases gradually for both spin channels
λ = 0, β = 0.65 1.66 2.321 391
λ = 0, β = 0.7143 1.64 2.310 380
as shown in Fig. 11. Although it is possible to select a param-
λ = 0.01, β = 0.65 4.66 2.259 486 eter set to produce a ground state electron density distribution
λ = 0.01, β = 0.7143 4.74 2.248 485 close to the KSDFT one, an accurate D0 or ωe would not be
λ = 0.01177, β = 0.65 4.82 2.260 512 guaranteed. This indicates that the current KEDF is still not
λ = 0.01177, β = 0.7143 4.91 2.249 472 general enough to predict good self-consistent energies and
electron densities simultaneously.

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-11 Orbital-free DFT for molecules J. Chem. Phys. 136, 084102 (2012)

TABLE X. Optimal parameter sets of OFDFT for all dimers with the LDA term proportional to 1/|r − r | or 1/q2 in the reciprocal space.
exchange-correlation functional.
Consequently, an approximate relation between λ and static
Dimer Best λ Best β
dielectric constant can be obtained:

Ms = 0 A
Al 0 0.51 λ∝ , (17)
ε̃(0) − 1
Ga 0 0.51
In 0 0.51
Si 0.01 0.65 where A is related to the kernel derivative, the density distri-
P 0.03 0.51 bution, and β. This equation implies that the larger the static
Sb 0.03 0.51 dielectric constant, the smaller the magnitude of λ. The sign
As 0.03 0.51 of A could not be universally determined, but in practice pos-
Li 0.01 0.7143 itive λ usually improves the results both for bulk materials
Ms = 1 and molecules. Physically, this relation also makes complete
Si 0.01 0.65
sense. For a metal with an infinitely large dielectric constant,
Al 0 0.7143
λ would be zero and the kernel would return to a similar form
Ga 0 0.7143
In 0 0.7143
as those in the WT or WGC KEDFs, whereas for a semicon-
ductor or insulator with a finite dielectric constant, λ would
be non-zero and consequently the gradient term would be ex-
plicitly included in the kernel.
Finally, reasonable parameters for different dimers are
For molecules, the static dielectric constant is not well
analyzed and suggested. Since satisfactory energy and den-
defined. However, the relationship derived above can be gen-
sity results cannot be guaranteed at the same time, here the
eralized at least conceptually. Usually in semiconductors or
parameters are tuned mainly based on D0 values. The opti-
insulators, the smaller the band gap is, the larger the static
mal parameter sets for each dimer are listed in Table X. In
dielectric constant is (as noted above, metals with zero band
order to analyze the trend in those parameters, we revisit the
gap have infinite dielectric constant). As a result, we may ob-
relationship between the parameter λ and the static dielectric
tain an approximate relation that the smaller the band gap,
constant.43 In the HC KEDF, if we recover the original non-
the smaller λ should be. In molecules, the band gap can be
local form of s(r),43
taken as the energy difference between the highest occupied
 
n(r) − n(r ) 1 molecular orbital (HOMO) and the lowest unoccupied molec-
s(r) = , (15) ular orbital (LUMO). Although strictly speaking the HOMO-
|r − r |
 n(r)4/3
LUMO energy gap in KSDFT is not equal to the true energy
and assume s(r) is very small so that we can perform a first- gap, it usually can give a rough estimate of the true gap. The
order Taylor expansion for the nonlocal term, then HOMO-LUMO gap energies from KSDFT calculations are
 shown in Table XI. For the group III family dimers, we know
TNL [n] = C n(r)8/3−β ω(kF (r)|r − r |)n(r )β drdr that the optimal λ is zero, and the HOMO-LUMO gap is the
  2 energy difference between the σ and π orbitals, which is very
n(r)−n(r ) small. For Si2 , the optimal λ equals 0.01, and the gap is the
+C n(r)8/3−β ω |kF |r−r | kF (r)λ
|r−r | energy difference between the π bonding and π * antibond-
|r − r |  β ing orbitals, which is significantly larger, as listed in the ta-
× n(r ) drdr . (16) ble. The situation is similar for the group V family dimers
n(r)8/3
with the optimal λ equal to 0.03, and even larger HOMO-
Inserting Eq. (16) together with the TF and vW terms into LUMO gaps. Thus the trend for the HOMO-LUMO gap is
Eq. (11) or Eq. (12) of Ref. 43 and carrying out the functional qualitatively consistent with the above relation. The smaller
derivative, we see that the TF and the vW KEDFs and the the HOMO-LUMO gap, the smaller λ is. In general, larger λ
first term on the right-hand side of Eq. (16) do not contribute is suggested to treat systems with large HOMO-LUMO gaps
to the 1/|r − r | or 1/q2 term, while one can show that the while zero or small λ is appropriate for systems with small
last term on the right-hand side of the Eq. (16) generates a HOMO-LUMO gaps.

TABLE XI. Energy differences ε between the HOMO and the LUMO in KSDFT calculations of each dimer at its re .
The LDA exchange-correlation functional is used.

Al2 Ga2 In2 Si2

Dimers (Ms = 0) (Ms = 0) (Ms = 0) (Ms = 0) P2 As2 Sb2 Li2

ε (eV) 0.21 0.21 0.11 2.50 3.70 3.33 2.37 1.59

Dimers Al2 Ga2 In2 Si2
(Ms = 1) (Ms = 1) (Ms = 1) (Ms = 1)
ε (eV) 0.20 0.23 0.13 2.37

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-12 Xia et al. J. Chem. Phys. 136, 084102 (2012)

V. CONCLUSIONS literature.21 For each element, target electron densities for fcc,
body-centered-cubic (bcc), simple cubic (sc), and CD crys-
In this work, a variety of diatomic molecules were cal-
tal structures are obtained by carrying out KSDFT calcula-
culated by OFDFT with the HC KEDF to test its ability to
tions with Troullier-Martins NLPSs (Ref. 18) generated by
describe covalent bonds in molecules as well as semiconduc-
the FHI98 code76 with default cutoff radii. The local term
tor crystals. By adjusting the two HC KEDF parameters, λ
is selected as the d channel for Li, Mg, Al, Si, P, As, and
and β, OFDFT can generate reasonably quantitative bond dis-
Sb, and the s channel for Ga and In. The inversion proce-
sociation energies, equilibrium bond lengths, and vibrational
dure described in Ref. 21 is carried out by a modified version
frequencies compared to KSDFT, a vast improvement over
of ABINIT.58 During the construction of the BLPSs, Fermi-
other KEDFs such as TFλvW or the WT KEDFs. However,
Dirac smearing with a smearing width equal to 0.1 eV is
for those dimers with triplet ground states, OFDFT predicts
used and the plane-wave basis kinetic energy cutoff is set to
the singlet state to have lower energy, which is not consistent
1600 eV. The k-point mesh used during the BLPS construc-
with KSDFT and experiment. This failure is further reflected
tion is 20 × 20 × 20 per unit cell for all structures. The num-
by comparing ground state electron densities. OFDFT yields
ber of atoms per unit cell is as follows: one atom for fcc, bcc,
a similar overall electron density distribution to KSDFT, but
sc, and body-centered tetragonal In (Ref. 78) structures, two
the density mismatch in the bonding region is still significant.
atoms for α-As,79 α-Sb,80 CD, and hexagonal closest-packed
Furthermore, OFDFT produces a significantly different spin
structures, three atoms for 9R Li,81 eight atoms for α-Ga,82
polarization distribution from the KSDFT one, which is likely
and A17 P.83 Equilibrium structures are relaxed with KSDFT
the reason for the incorrect energy ordering between differ-
with a force threshold of 5 × 10−5 hartree/bohr and a stress
ent spin states. Finally, the effects of parameters on proper-
threshold of 5 × 10−7 hartree/bohr.3 The structures are then
ties were analyzed. In terms of absolute energies, larger λ and
expanded and compressed by 2% to obtain total energy ver-
smaller β tend to raise the total energy, but the relative trend
sus volume points, which are fit to Murnaghan’s equation of
of total energy for each dimer at its equilibrium bond length is
state77 to calculate bulk moduli. The total energy difference
similar and consistent with KSDFT results. The λ value also
between different phases at their equilibrium volumes is the
influences the density distribution. Increasing λ tends to lower
phase energy difference.
the electron density in the bonding region. Although by tuning
Two parameters in the BLPS construction, the value of
λ it is possible to make the OFDFT density match reasonably
the non-Coulombic part of the BLPS at q = 0 in reciprocal
well with the KSDFT one, the current KEDF cannot guar-
space and the position beyond which the BLPS recovers the
antee a good electron density and bond dissociation energy at
Coulomb tail in real space, are adjusted so as to reproduce
the same time. Unlike the equilibrium bond length which does
the equilibrium volumes and bulk moduli calculated by
not change much when adjusting parameters, the bond disso-
KSDFT with the NLPS, for the experimental ground state
ciation energy is very sensitive to the λ value. Zero and small
phase of each element. The BLPSs are further tested for
λ generates shallow wells and small D0 values, while large λ
each element with different crystal structures other than its
predicts deep wells and large D0 values. Finally, the optimal
ground state phase. In the test calculations, 900 eV kinetic
value of λ was qualitatively related to a molecule’s HOMO-
energy cutoffs and 20 × 20 × 20 k-point meshes in each unit
LUMO gap. Zero or small λ is suggested for molecules with
cell are employed. Fermi-Dirac smearing with a smearing
small HOMO-LUMO gaps and large λ is better for molecules
width of 0.1 eV is used for metallic solids and no smearing
with large HOMO-LUMO gaps.
is used for insulators. Comparisons of results obtained using
Although the HC KEDF can predict quite good results
the NLPSs versus the BLPSs are listed in the supplemental
for D0 , re , and ωe for diatomic molecules, it still produces
material Tables S1-S9.84 The new BLPSs are also plotted in
considerable errors, such as significant inaccuracy in describ-
the supplemental information Figures S1-S3.84 The BLPS
ing spin state ordering and the related mismatch in the elec-
results are in good agreement with the NLPS results overall.
tron spin densities. Further research is needed to improve the
Except for the CD structure of Li, the BLPS bulk moduli
KEDF and potentially the LPS to make OFDFT a completely
are generally reasonably close to the NLPS values. For equi-
reliable and general quantum mechanics simulation tool.
librium volumes, the difference between BLPS and NLPS
predictions is quite small for all structures and elements.
ACKNOWLEDGMENTS Finally, the energy orderings predicted by BLPSs exactly
We are grateful for support from the Office of Naval Re- reproduce the NLPS results, and the phase energy differences
search and the National Science Foundation (NSF). also generally agree well with the NLPS results, except in
some cases such as fcc Si. The ground state of Ga is predicted
by the NLPS and the BLPS to be fcc instead of α-Ga, which
APPENDIX: CONSTRUCTION AND TESTING
may be due to the lack of a nonlinear core correction.85 The
OF NEW BLPSs
fcc structure is predicted to be the ground state of Li, but the
For both KSDFT and OFDFT calculations, BLPSs energy differences between fcc, bcc, and the experimentally
(Ref. 21) are employed with the PZ (Refs. 49 and 50) LDA observed 9R structure are very small, less than 2 meV, which
and the PBE (Ref. 53) GGA exchange-correlation function- is certainly within the KSDFT LDA or GGA uncertainty.
als. For all LDA BLPSs except for Li, we use previously re-
ported ones.21, 43 The Li LDA BLPS and all GGA BLPSs are 1 P. Hohenberg and W. Kohn, Phys. Rev. 136, B864 (1964).
2 W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 (1965).
constructed with the same inversion method described in the

Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
084102-13 Orbital-free DFT for molecules
3 W. Kohn, Phys. Rev. Lett. 76, 3168 (1996).
4 S. Goedecker, Rev. Mod. Phys. 71, 1085 (1999).
5 S. Ismail-Beigi and T. A. Arias, Phys. Rev. Lett. 82, 2127 (1999).
6 S. Y. Wu and C. S. Jayanthi, Phys. Rep. 358, 1 (2002).
7 T. A. Arias, Rev. Mod. Phys. 71, 267 (1999).
8 F. Shimojo, R. K. Kalia, A. Nakano, and P. Vashishta, Phys. Rev. B 77,
085103 (2008).
9 Y. A. Wang and E. A. Carter, in Theoretical Methods in Condensed Phase
Chemistry, edited by S. D. Schwartz (Kluwer, Dordrecht, 2000), p. 117.
10 L. Hung and E. A. Carter, Chem. Phys. Lett. 475, 163 (2009).
11 I. Shin, A. Ramasubramaniam, C. Huang, L. Hung, and E. A. Carter,
Philos. Mag. 89, 3195 (2009).
12 Q. Peng, X. Zhang, L. Hung, E. A. Carter, and G. Lu, Phys. Rev. B 78,
054118 (2008).
13 L. Hung and E. A. Carter, J. Phys. Chem. 115, 6269 (2011).
14 L. Hung and E. A. Carter, Modell. Simul. Mater. Sci. Eng. 19, 045002
(2011).
15 I. Shin and E. A. Carter, Modell. Simul. Mater. Sci. Eng. 20, 015006
(2011).
16 D. R. Hamann, M. Schlüter and C. Chang, Phys. Rev. Lett. 43, 1494 (1979).
17 L. Kleinman and D. M. Bylander, Phys. Rev. Lett. 48, 1425 (1982).
18 N. Troullier and J. L. Martins, Phys. Rev. B 43, 1993 (1991).
19 B. Zhou, Y. A. Wang, and E. A. Carter, Phys. Rev. B 69, 125109 (2004).
20 S. Watson, B. J. Jesson, E. A. Carter, and P. A. Madden, Europhys. Lett.
41, 37 (1998).
21 C. Huang and E. A. Carter, Phys. Chem. Chem. Phys. 10, 7109 (2008).
22 B. Zhou and E. A. Carter, J. Chem. Phys. 122, 184108 (2005).
23 L. H. Thomas, Proc. Cambridge Philos. Soc. 23, 542 (1927).
24 E. Fermi, Rend. Accad. Naz. Lincei 6, 602 (1927).
25 E. Fermi, Z. Phys. 48, 73 (1928).
26 E. Teller, Rev. Mod. Phys. 34, 627 (1962).
27 E. H. Lieb and B. Simon, Adv. Math. 23, 22 (1977).
28 C. F. v. Weizsäcker, Z. Phys. 96, 431 (1935).
29 K. Yonei and Y. Tomishima, J. Phys. Soc. Jpn. 20, 1051 (1965).
30 M. Levy, J. P. Perdew and V. Sahni, Phys. Rev. A 30, 2745 (1984).
31 C. H. Hodges, Can. J. Phys. 51, 1428 (1973).
32 D. R. Murphy, Phys. Rev. A 24, 1682 (1981).
33 E. Chacón, J. E. Alvarellos, and P. Tarazona, Phys. Rev. B 32, 7868 (1985).
34 P. García-González, J. E. Alvarellos, and E. Chacón, Phys. Rev. B 53, 9509
(1996).
35 P. García-González, J. E. Alvarellos, and E. Chacón, Phys. Rev. B 57, 4857
(1998).
36 L. W. Wang and M. P. Teter, Phys. Rev. B 45, 13196 (1992).
37 Y. A. Wang, N. Govind, and E. A. Carter, Phys. Rev. B 60, 16350 (1999);
64(E), 089903 (2001).
38 E. Smargiassi and P. A. Madden, Phys. Rev. B 51, 117 (1995).
39 K. M. Carling and E. A. Carter, Modell. Simul. Mater. Sci. Eng. 11, 339
(2003).
40 G. Ho, M. T. Ong, K. J. Caspersen, and E. A. Carter, Phys. Chem. Chem.
Phys. 9, 4951 (2007).
41 B. Zhou, V. L. Ligneres, and E. A. Carter, J. Chem. Phys. 122, 044103
(2005).
42 R. M. Pick, M. H. Cohen, and R. M. Martin, Phys. Rev. B 1, 910 (1970).
43 C. Huang and E. A. Carter, Phys. Rev. B 81, 045206 (2010).
44 K. Yonei, J. Phys. Soc. Jpn. 31, 882 (1971).
45 J. P. Perdew, M. Levy, G. S. Painter, S. Wei, and J. B. Lagowski, Phys. Rev.
B 37, 838 (1988).
46 S. Iyengar, M. Ernzerhof, S. N. Maximoff, and G. E. Scuseria, Phys. Rev.
A 63, 052508 (2001).
Downloaded 31 May 2012 to 134.245.41.90. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
J. Chem. Phys. 136, 084102 (2012)
47 L. A. Constantin and A. Ruzsinszky, Phys. Rev. B 79, 115117 (2009).
48 V. Cocula, C. J. Pickard, and E. A. Carter, J. Chem. Phys. 123, 214101
(2005).
49 D. M. Ceperley and B. J. Alder, Phys. Rev. Lett. 45, 566 (1980).
50 J. P. Perdew and A. Zunger, Phys. Rev. B 23, 5048 (1981).
51 J. P. Perdew and Y. Wang, Phys. Rev. B 45, 13244 (1992).
52 U. von Barth and L. Hedin,J. Phys. C 5, 1629 (1972).
53 J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865
(1996).
54 G. L. Oliver and J. P. Perdew, Phys. Rev. A 20, 397 (1979).
55 H. Jiang and W. Yang, J. Chem. Phys. 121, 2030 (2004).
56 G. Ho, V. L. Ligneres, and E. A. Carter, Comput. Phys. Commun. 179, 839
(2008).
57 L. Hung, C. Huang, I. Shin, G. Ho, V. L. Ligneres, and E. A. Carter, Com-
put. Phys. Commun. 181, 2208 (2010).
58 X. Gonze, J. M. Beuken, R. Caracas, F. Detraux, M. Fuchs, G. M.
Rignanese, L. Sindic, M. Verstraete, G. Zerah, F. Jollet, M. Torrent, A. Roy,
M. Mikami, P. Ghosez, J.-Y. Raty, and D. C. Allan, Comput. Mater. Sci. 25,
478 (2002).
59 Y. Tomishima and K. Yonei, J. Phys. Soc. Jpn. 21, 142 (1966).
60 D. S. Ginter, M. L. Ginter, and K. K. Innes, Astrophys. J. 139, 365 (1964).
61 T. H. Upton, J. Phys. Chem. 90, 754 (1986).
62 C. W. Bauschlicher, H. Partridge, S. R. Langhoff, P. R. Taylor, and
S. P. Walch, J. Chem. Phys. 86, 7007 (1987).
63 M. F. Cai, T. P. Dzugan, and V. E. Bondybey, Chem. Phys. Lett. 155, 430
(1989).
64 H. J. Himmel and B. Gaertner, Chem.-Eur. J. 10, 5936 (2004).
65 X. Tan and P. J. Dagdigian, J. Phys. Chem. A 107, 2642 (2003).
66 B. Song and P. Cao, J. Chem. Phys. 123, 144312 (2005).
67 K. K Das, J. Phys. B 30, 803 (1997).
68 Y. Zhao, W. Xu, Q. Li, Y. Xie, and H. F. Schaefer III, J. Phys. Chem. A
108, 7448 (2004).
69 G. Balducci, G. Gigli, and G. Meloni, J. Chem. Phys. 109, 4384 (1998).
70 K. Balasubramanian and J. Li, J. Chem. Phys. 88, 4979 (1988).
71 K. Huber and G. Herzberg, Constants of Diatomic Molecule (Van Nostrand
Reinhold, New York, 1979).
72 F. W. Froben, W. Schulze, and U. Kloss, Chem. Phys. Lett. 99, 500
(1983).
73 R. G. Parr and W. Yang, Density Functional Theory of Atoms and
Molecules (Oxford University Press, New York, 1989).
74 H. Gomez, T. R. Taylor, Y. Zhao, and D. M. Neumark, J. Chem. Phys. 117,
8644 (2002).
75 Z. Gan, D. J. Grant, R. J. Harrison, and D. A. Dixon, J. Chem. Phys. 125,
124311 (2006).
76 M. Fuchs and M. Scheffler, Comput. Phys. Commun. 119, 67 (1999).
77 F. D. Murnaghan, Proc. Natl. Acad. Sci. U.S.A. 30, 244 (1944).
78 Smithells Metals Reference Book, edited by E. A. Brandes and G. B. Brook,
7th ed. (Elsevier, New York, 1998).
79 L. F. Mattheiss, D. R. Hamann, and W. Weber, Phys. Rev. B 34, 2190
(1986).
80 L. M. Falicov and P. J. Lin, Phys. Rev. 141, 562 (1966).
81 A. W. Overhauser, Phys. Rev. Lett. 53, 64 (1984).
82 M. Bernasconi, G. L. Chiarotti, and E. Tosatti, Phys. Rev. B 52, 9988
(1995).
83 A. Brown and S. Rundqvist, Acta Crystallogr. 19, 684 (1965).
84 See supplementary material at http://dx.doi.org/10.1063/1.3685604 for the
comparison of bulk properties predicted by BLPSs and NLPSs for each
element, as well as plots of the newly constructed BLPSs.
85 S. G. Louie, S. Froyen, and M. L. Cohen, Phys. Rev. B 26, 1738 (1982).