You are on page 1of 17

Hardness and Softness

in Density Functional Theory


Jos~ L. Gdzquez
Dcpartamento de Quimica, Divisi6n de Ciencias Bfisicas e Ingenieria, Universidad Autrnoma
Metropolitana-lztapalapa, A.P. 55-534, M~xico, D.F. 09340, Mexico

The fundamental equations to describe the change from one ground-state to another, in the
framework of density functional theory, are used to analyze a set of hardness and softness functions
that are hierarchized as non-local, local and global quantities. Through these definitions it is shown
that under conditions of constant chemical potential, the interaction between two chemical systems
evolves towards a state of maximum hardness, and that soft-soft, and hard-hard interactions are
energetically favored. It is also shown that to a good approximation, the ground-state energy of
a system decreases when its hardness increases. Possible applications of these principles are briefly
discussed.

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2 Fundamental Equations for the Energy Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 Local Hardness and Global Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4 Chemical Potential, Hardness, Ionization Potential, and Electron Affinity . . . . . . . . . . . 33
5 The Interaction Energy and the Principle of Maximum Hardness . . . . . . . . . . . . . . . . . 35
6 The Principle of Hard and Soft.Acids and Bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7 Scaling Properties of Simple Functionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
8 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Structure and Bonding, Vol. 80


f~.~Springer-Verlag Berlin Heidelberg 1993
28 Jos6 L. Gfizquez

1 Introduction

The principle of hard and soft acids and bases [I] (HSAB), and the principle of
electronegativity equalization 1,2], together with frontier orbital theory 1,3] have
been, over the years, very useful to establish the behavior of molecules under
different circumstances, their reactive sites, and possible reaction mechanisms
[4]. Through these principles, and through the values of the parameters asso-
ciated with them: hardness, softness, and electronegativity, it has been possible
to correlate and to analyze experimental information that allows one to charac-
terize the interactions involved between different chemical species in many
different situations. From this experience it has been possible to establish
a priori the development of a wide variety of chemical reactions.
Recently, the maximum hardness principle I-5-9] has been added to the list
of practical concepts, and it is expected to become a powerful tool in the analysis
of chemical behavior.
In the last few years, all these concepts have been found to be intimately
related with fundamental variables of density functional theory [10]. This
situation has provided a solid theoretical basis to the principles just mentioned,
and has led to operational formulas that allow one to quantify the associated
parameters. In addition, through density functional theory it has been possible
to build a bridge between these rather intuitive concepts, that provide a frame-
work for simple physical interpretations of complex phenomena, and wavefunc-
tion theory, that provides an accurate description of the electronic structure of
chemical systems, but otherwise far from providing a framework for simple
interpretations. In brief, density functional theory is able to take the relevant
information contained in the wavefunction, and transform it into an almost
pictorial representation, ready to be analyzed through the principles just men-
tioned above [10-22].
The purpose of this work is to start from the basic equations of density
functional theory to describe the changes in the energy associated with the
transition from one ground-state to another, in terms of different sets of
variables. In this process one will find the natural definitions of the hardness and
softness kernels, the local hardness, the local softness, the global hardness and
the global softness 123]. Then, we will proceed to establish their relation with
ionization potentials and electron affinities, in order to confirm their behavior as
a measure of chemical hardness or softness 1-14, 24]. Finally, this theoretical
framework will be used to analyze the maximum hardness and the HSAB
principles.
Hardness and Softness in Density Functional Theory 29

2 Fundamental Equations for the Energy Changes

In density functional theory, the total energy of an atom or a molecule in terms


of its electronic density p(r) is written as 1,25,1
E[p] = S drv(r)p(r) + F [ p l (1)
where v(r) is the external potential generated by the nuclei, and F1,p-I is the
universal Hohenberg-Kohn functional composed by the electronic kinetic
energy, and the electron-electron interaction energy. The minimization of the
energy with respect to the electronic density, under the constraint that the
integral of the latter over the whole space must be equal to the number of
electrons N, leads to the Euler-Lagrange equation of the form
6F
Ix = v(r) + - (2)
So(r)
where Ix (the Lagrange multiplier) is the chemical potential of the system 1,11,1.
Now consider the change from one ground-state to another. One ground-
state will be characterized by p~ v~ (r), Ix~ N ~ and E ~ while the other one will
be defined by p(r), v(r), IX, N, and E. The difference between the two Euler-
Lagrange equations is given by

AIX= 8v(r) + (3)


66--~p(~) 2p-~r)po(,,
where AB = IX - Ixo, and 6v(r) = v(r) - v~
By performing a Taylor series functional expansion of one of the functional
derivatives in Eq. (3), around the other one, and keeping only terms up to first
order, one may rewrite this equation in the form
8v(r) = AIX - S dr'rl ~ r') ~Sp(r') (4)
where 5p(r') = p(r') - p~ and 1,23,1
82F
rl(r, r') - 6p(r')~Sp(r) (5)

is the hardness kernel. The superscript "o" in Eq. (4) appears to indicate that the
second functional derivative is evaluated at p~ The softness kernel s(r, r') is
defined as the inverse of the hardness kernel 1-23-1, thus
]"dr"s(r, r")rl(r", r') = 8(r - r') (6)
This expression may be combined with Eq. (4) to obtain
8p(r) = s ~(r) Apt -- ~ dr" s~ (r, r") ~v (r") (7)
where s(r) is the local softness 1,26,1, and it is defined by 1,23,1
s(r) = Sdr's(r,r') (8)
30 Jos6 L. Gfi.zquez

Integrating Eq. (7) over r one finds that


AN = s~ - ~ drs~ (9)
where s ~ is the global softness [26], and it is given by [23]
s -- ~S dr dr's(r, r') -- ~ dr s(r) (I0)
From Eq. (9), if AN = 0, then
Ag = ~ dr f(r) 5v (r) (11)
where
s(r)
f(r) = - - (12)
s
is the Fukui function [271, a quantity closely related to the frontier orbitals.
Thus far, one can see that the definitions of softness functions appear in
a natural way when the difference between two Euler-Lagrange equations is
considered. Local hardness [281 and global hardness [12, 281 will be discussed
in the next section to complete the set of hardness functions.
The local relations between ~p(r) and 5v(r), Eqs. (4) and (7) together with the
global relation between AN and ~v(r), Eq. (9), are very important to study
energy changes up to second order associated with the transition from one
ground-state to another. Through the use of Eq. (4) one can express the changes
in the external potential in terms of the changes in the chemical potential and the
changes in the electronic density, while in Eq. (7) the roles of ~p(r) and ~v(r) are
inverted. The coefficients that measure the ability of the system to undergo
certain changes are precisely the hardness and softness functions.
Now let us focus on the energy difference between the two ground-states
considered. Using Eq. (1), one has that
AE = E[O] - E[O~
= S drv(r)p(r) - S dr v" (r) p~ (r) + F [ p ] - F[p'~] (13)
By performing a Taylor series functional expansion of F [ p ] around F [ p ~ and
retaining terms up to second order, one can express Eq. (13) in the form
AE = B~ + .[ dr[p~ + ~p(r)]~v(r)
+ 89~ drdr' rl~ r') ~p(r) ~p(r') (14)
This expression, as written, depends on tip(r) and ~v(r), AN is just the integral of
8p(r) over all space. However, in order to analyze energy changes under specific
circumstances it is convenient to express it in terms of other sets of variables.
Through the use of Eqs. (4), (7), and (9) one can transform Eq. (14) into
expressions for AE in terms of the sets: {AN, Sv(r)}, {Ag, fv(r)}, and
{Ala, 8p(r)}. The equations thus obtained are given in Table 1, and they will be
used throughout this work to describe different processes involved in different
interactions between chemical species.
Hardness and Softness in Density Functional Theory 31

Table 1. Energy changes in terms of different sets of variables

Expressions for the set {AN, 8v(r)}:

Ap = I A N + ~ dr ss~(f) 8v(r) (T-l)

aptr) = s~(r)ANs
~ +J"dr' [L s~176
~ s~ r')JSv(r')- (T-2)

i 1 2
AE = labAN + Jdr p~(r);Sv(r) + 2 ~(AN) + AN~dr--~-Sv(r)
s~
(T-3)
+ f l'drdr'[ -s~ s~
"" L s~ /
Expressions for the set {Al.t, By(r)}:
AN = s~ - ~dr s~(r)8v(r) (T-4)

8p(r) = s~(r)Al~ -- ~dr's~ r')Sv(r') (T-5)


AE = p.~'s~ + ~dr[p~ - la~176 8v(r)
+ 89176 2 - 89 dr's~ r')Sv(r)Sv(r') (7-6)
Expressions for the set {Art, 8p(r)}:
AN = ~drSp(r) (T-7)
By(r) = Ala - ~dr'~~ r')Sp(r') (T-8)
AE = p~ + ApAN + N"Ala
- 89J'~drdr'q~ r')[2 p~ + 8p(r)] 8p(r') (T-9)

3 L o c a l H a r d n e s s and G l o b a l H a r d n e s s

Just as in the case of the softness functions, where integration of the kernel leads
to local and global quantities, one may introduce local and global hardnesses
through integrals of rl(r, r'). Thus, the local hardness has been defined as [28]

rl(r) = S dr'rl(r, r') p(r') (15)


N
and the global hardness has been found to be given by 1-23, 28]
q = S dr f(r) q (r) (16)
These expressions fulfill the condition of being, in each case, the inverse of the
corresponding softness function. That is [23],
~drs(r)rl(r) = 1 (17)
and [23]
srl = 1 (18)
32 Jos6 L. G~.zquez

However, one does not need to define a local hardness in order to define the
global hardness as the inverse of the global softness, nor does one have to define
q(r) as in Eq. (15), in order to obtain an inverse relation as the one given in
Eq. (16). This can be proved by noting that the integral of Eq. (6) over r leads to
1
S dr"f(r")rl(r", r') = - (19)
s

Note that the right hand side of this equation is independent oft'. Thus, on one
hand, for any function a(r) such that j'dra(r) = 1, one may define
q~ (r) = I dr' rl (r, r') cc(r') (20)
This definition of local hardness satisfies Eqs. (16) and (17). On the other hand, it
is clear from Eq. (19) that in the approach followed in the previous section, the
global hardness is determined by the inverse of s, and does not arise as an
independent concept, in contrast to the global softness.
Of course, if one considers the derivatives of the energy with respect to the
number of electrons, one finds that [12]

n = = , (21)

Wtiile the local softness is given by [26]

s(r) = (Sp(r)~ (~p(r)'~ ( ~ N ) (22)


\ i~IX) , = t O N , / , , ~ ,,
where [27]

f(r) = /)\/~P(r'/ (23)


\aN/,
and [26]

s = (24)
V

In this framework, q and s appear through independent interpretations, but


they turn out to be the inverse of each other.
However, in this context the concept of local hardness may be ambiguous in
the sense expressed by Eq. (20). This can be seen by calculating the hardness of
Eq. (21) by direct differentiation of the energy functional with respect to N.
Starting from Eq. (1) one may prove that [29]

n -- (~ZE'~
koN ) = I d r r ( r ) I d r ' f ( r ' ) n ( r , r') (25)

But if the energy is written as [11, 30]

5F (26)
E [ p ] = Nla + F [ p ] - I d r p ( r ) s p ( r )
Hardness and Softness in Density Functional Theory 33

then one may prove that the direct differentiation with respect to N leads to [29.]

q = = j*drf(r)Sdr' rl(r , r') (27)


v

Equations (25) and (27) lead to the same value of rl, but they imply different
definitions of the local hardness. This behavior is in agreement with Eqs. (19)
and (20): Eq. (25) corresponds to a ( r ) = f(r), while Eq. (27) corresponds to
a(r) = p(r)/N, the latter is the original definition of local hardness [28].
It is interesting to analyze the long range behavior of the generalized local
hardness defined by Eq. (20). Because of the exponential decay of the electronic
density, one can show that far away from the nuclei, the dominant term comes
from the coulombic contribution to the hardness kernel [281, that is

rh(r) ~ :f d r ' - Ir
~ - '-)
r'l (28)

Thus, one can see that the original definition of local hardness, with
a(r) = p(r)/N nicely embodies electrostatic effects. Far away from the nuelei the
local hardness becomes proportional to the electrostatic potential generated by
the molecular charge distribution. Since contour diagrams of molecular elec-
trostatic potentials have been widely used to analyze the chemical reactivity of
a great number of chemical species [31-1, it is to be expected that the local
hardness index provided by Eq. (15) will incorporate additional effects, and this
way it may become a very useful reactivity index.
On the other hand, if c~(r) = f(r) the local hardness becomes proportional to
the electrostatic potential generated by the molecular Fukui distribution. Since
far away from the nuclei the electronic density will be practically equal to the
density of the highest occupied molecular orbital PuoMo(r), and since the Fukui
function is closely related to PHOMO(r),one should expect that in this limit both
definitions will lead to very similar results. However, at intermediate distances
from, and close to, the nuclei they may provide different information on the
reactive sites of a chemical species.
In summary, one can see that density functional theory provides a coherent
quantitative language of hardness and softness functions that may be classified
as non-local, local and global reactivity indexes.

4 Chemical Potential, Hardness, Ionization Potential,


and Electron Affinity

The expressions derived in Sect. 2 will be used now to express some quantities in
terms of the first ionization potential I and the electron affinity A.
Consider the case in which the two ground-states are a neutral atom or
molecule (with N electrons), and its corresponding positive ion (with N - 1
34 Jos6 L. G;izquez

electrons). If the two electronic densities are determined for the same geometric
structure 15v(r) = 0 for all values of r, A~t = IXN-t -- lX~, and ~Sp(r) = pN-l(r)
- pN(r). Then one can prove that [15, 17, 32]
f - (r) ~ PN(r) -- PN- 1(r) (29)
and [32]
I = EN-I -- EN
= -- IX~ + 89 dr'rl~ r')(pN_l (r) -- pf~(r))(pN-1 (r') -- p?~(r'))
(30)
Similarly, in the case of the neutral species, and its corresponding negative ion
(with N + 1 electrons), one can show that [15, 17, 32]
f+(r) ~ p N + l ( r ) - pN(r) (31)
and 1-32]

A = EN - E~+I
= - IX~ + 89j ~ d r d r ' r l ~ r')(pN+l(r) -- p ~ ( r ) ) ( p N + l ( r ' ) - p~(r'))
(32)

Through the use of these relations and Eq. (19), one finds that
I+A
go ~ (33)
2
and
rl ~ = ~Sdrdr'f(r)f(r')rl~ r') ~ I - A (34)
Equations (29) and (31) give the left and right approximations to the Fukui
function, while Eqs. (33) and (34) represent the average of the left and right
approximations. They all correspond to the correct expressions when one
retains terms up to second order in the energy [17]. In fact, Eqs. (33) and (34)
provide very strong support to the interpretations of IX as a measure of elec-
tronegativity, and to 1] as a measure of hardness. Using experimental values of
I and A one finds that Ix and 1] are in good agreement with respect to chemical
behavior, both as to acid-base character, and as to chemical hardness [14, 243 .
One can find alternative expressions if the relaxation effects associated with
the addition or removal of charge are neglected. By holding the electronic
charge distribution of the neutral system constant, one can find several different
expressions depending on how the process of addition or removal is carried out.
In a molecular orbital framework it seems natural to remove charge from the
highest occupied molecular orbital, and to add charge into the lowest
unoccupied molecular orbital [27, 33]. In this case,
f - ( r ) ~ PUOMO(r) (35)
Hardness and Softness in Density Functional Theory 35

and
f+ (r) ,~ PLUMp(r) (36)
Equations (35) and (36) show that the Fukui function embodies the frontier
orbitals.
However, in a density functional framework it seems natural to remove and
to add charge uniformly. Thus in this case
N-1
PN- 1(r) ~, ~ pN(r) (37)

and
N+I
p~+l(r) ~ ~ p N ( r ) (38)

These approximations lead to

f _ (r) ~ f+(r) ~ --N--


p ~(r) (39)

While substituting Eq. (37) in (30) and Eq. (38) in (32) one finds that

SSdr dr' p(r)


N p(r')
N rl "r,
( r')~I-A~r I (40)

where the second equality follows from Eq. (34). This expression for the
global hardness will be used in the next section.
It is important to mention that the three sets of approximations to the Fukui
function (Eqs. (29), (31)), (Eqs. (35), (36)), and (Eqs. (39)) lead to Eq. (33), but they
lead to different representations of the global hardness. Which one leads to the
best description of the experimental values of the quantity (I - A) would have to
be proved numerically. However, since the experimental values of the quantity
(I - A) do conform with the chemical hardness concept l'5J, one may conclude
that the three representations may be used almost indistinctly to describe the
global hardness of a chemical species.

5 The Interaction Energy and the Principle o f M a x i m u m Hardness

The previous sections were devoted to the description of the parameters asso-
ciated with the inherent chemical reactivity of molecules. N o w we will proceed
to analyze the principles governing the evolution of these parameters when
isolated species come into interaction.
Thus one can see from Eq. (33) that the electronegativity is just the negative
of the chemical potential of density functional theory [11]. Through this
36 Jos6 L. G~izquez

identification, the principle of electronegativity equalization can be readily


understood, since the chemical potential (electronegativity) of the system must
be uniform throughout the space occupied by the molecule. When two molecu-
les come into interaction, then some charge is transferred from one of them to
the other in order to equalize their chemical potentials (electronegativities).
The principle of HSAB, and the principle of maximum hardness need more
elaboration in order to be theoretically justified. The latter was originally stated
by Pearson, who concluded that [51 "there seems to be a rule of nature that
molelcules arrange themselves so as to be as hard as possible". Later on, Zhou
and Parr [6-8"1 found some evidence about the validity of this principle under
conditions of constant temperature and chemical potential, and recently Parr
and Chattaraj l-9] have provided a formal proof using statistical mechanics.
Here this principle will be analyzed from the point of view of the changes that
take place in the electronic structure of two systems that come into interaction.
Consider the general case
A + B ~ AB (41)
that is, A (an atom or a molecule) interacts with B to give a third species AB. The
interaction energy is given by

AEi,,t = E[pAB] - - E[p,~] - E[p~] (42)


where pAB(r) is the electronic density of the system AB at equilibrium (minimum
energy), and p~,(r) and pg(r) are the electronic densities of the isolated systems.
Now, consider that the interaction can be divided into two steps. In the first
step, when A and B are located far apart from each other, their chemical
potentials, ~t~, and la~, change to reach a common value rtA.. This step occurs
through charge transfer at constant external potential, and it will be assumed
that ~tA, is equal to the chemical potential of the system AB at equilibrium. Thus
in this case the energy change can be written as
AE, = AE$ + AE~ (43)
where
AEvA = E [ p * ] - E[p~] (44)
and
AE~ = E[p~] - E[p~] (45)
Here p*(r) corresponds to system A with v,~(r) and laAa, while p*(r) corresponds
to system B with vg(r) and I~A,.
In the second step, A and B evolve towards the equilibrium state through
changes in the electronic density of the global system AB produced by changes
in the external potential VAR(r). This step occurs under conditions of constant
chemical potential, and can be expressed in the form
AE~ = E[pA~] -- E [P~.B] (46)
Hardness and Softness in Density Functional Theory 37

where p~a(r) = p*(r) + p*(r) is the electronic density of the system AB with
~tAa when A and B are far away from each other. Using Eqs. (43)-(46), and
noting that
E[O%a] = E[p~] + E[p~] (47)
and finds that Eq. (42) can be rewritten in the form

AEI,t = AEv + AE, (48)

In order to evaluate each contribution in terms of the reactivity parameters


one can make use of the relations given in Table 1. Thus, for AE~ and AE~ it is
convenient to use the set {AI.t, 6v(r)}, with 5v(r) = 0 for all values ofr (Eq. (T-6)).
If the reference electron densities are taken to be p*(r) and p~(r) in each case,
one finds
AEv = - - l2.SA (~AB - - l'tA) 2 - - 12 .SB (~AB _
~.1~)2 (49)
where the global charge conservation has been used.
In the case of AE~ it is convenient to use Eq. (T-9) of the set {Ala, ~Sp(r)}, with
Art = 0 and AN = 0. If the reference electron density is taken to be pAn(r), then
AE~ = - 89S~drdr' qAB(r, r')pAn(r)pAB(r')
+ 89Sj'dr dr' rl~a (r, r') p~a (r) p~n (r') (50)
where rlAa(r, r') is evaluated at PAn(r), and rl*a(r, r') which is evaluated at p*B(r),
has replaced rlAa(r, r') in the second integral, because they are equal to each
other if only the zeroth-order term of a Taylor series functional expansion of the
latter around the former is kept. This level of approximation in the second
functional derivatives is consistent with the second order approximations used
to describe the energy changes. The two integrals of Eq. (50) may be written in
terms of the global hardness by using Eq. (40). Therefore
AE~ = _ 12 NAB
2 (1"lAB_ l ] ~ n ) (51)

This relation leads to the principle of maximum hardness, because if one


assumes that the interaction between A and B leads to an equilibrium state of
lower energy than the total energy of A and B when they are far apart from each
other, then rlAa > rl*n, AE~ < 0. The state of maximum hardness can be
associated to the state of minimum energy. On the other hand, if the equilibrium
state of AB occurs when A and B are very far apart, then it means that
q~,a > qAa which also implies a state of maximum hardness. All together,
Eq. (51) establishes that under conditions of constant chemical potential, the
system AB evolves towards a state of maximum hardness.
The set of Eqs. (49) and (51) provide a very interesting interpretation about
the interaction between two species. Since AEv < 0, it implies that the lowest
value will be reached for the greatest possible values of s~, and s~. Thus, in the
first step (Eq. (49)) A and B evolve to become as soft as possible, while in the
second step (Eq. (51)) they evolve as a whole towards an equilibrium state AB of
38 Jos6 L. G~izqucz

maximum hardness. This result agrees with experimental evidence: soft molecu-
les are more reactive than hard molecules [5, 7, 8].

6 The Principle o f Hard and Soft Acids and Bases

Starting from the analysis presented in the previous section we will now try to
understand why "hard likes hard", and "soft likes soft" (HSAB principle) [1, 34].
In this case it is convenient to express Eqs. (49) and (51) in terms of the initial
values of the parameters, rather than in terms of the final values of the para-
meters.
In the case of Eq. (49) this means that the reference electron density in
Eq. (T-6) is p~(r) instead of p*(r), and similarly for B, then
AE, ~_
2 SA(la~n
1 o --
~t~2) + l2o~Bt~,An
o1,,2
- Ix~2) (52)
Since laAn has been defined as the chemical potential of AB at equilibrium, it is
composed of two contributions: The first one comes from charge transfer, while
the second one comes from the change in the external potential. If the latter is
neglected, then Eq. (52) can be written in the form [12, 35]

1 (rt~ - ~t~)2
AE, ~ 2 rl~, + rl~ (53)

which is given in terms of the initial values of the reactivity indexes. Since
AEv < 0, for a given chemical potential difference, the stabilization is larger
when A and B are soft.
In the case of Eq. (51) one can make use of the approximate additivity of the
softness of the constitutive parts [18, 36] to evaluate the softness of the system
AB. In general,

SAB = k'(s~ + s~) (54)

where k' is a proportionality constant. Ifk' = 1/2, Eq. (53) becomes the arithme-
tic average, which has been found to give a fairly good estimate of the global
softness of a system in its equilibrium state, in terms of the global softness of the
constitutive parts 1,36].
On the other hand, when A and B are far apart from each other 1-18], one
would expect the total softness to be roughly equal to s* + s~, which corres-
ponds to k ' = I. However, since the global softness is quite insensitive to the
number of electrons [37], one may consider the softness values of the constitut-
ive parts either before or after the charge transfer has occurred [ 3 5 1 and
therefore s~,a may be approximated through Eq. (54) with a different propor-
tionality constant, k". This way Eq. (51) can be expressed in terms of the initial
Hardness and Softness in Density Functional Theory 39

values of the reactivity indexes in the form


1
AE. ~ -- 12 N A
2 Bk~ (55)
s~, + s~,
Since AE, < 0, the stabilization is larger when A and B are hard.
Equations (53) and (55) show opposite behavior with respect to the relative
values o f q ~ for a given rl~,. In Eq. (53), for a given q~. the smaller rl~ the better,
while in Eq. (55) for a given rl~,, the larger rl~ the better. Since the sum of
Eqs. (53) and (55) leads to the total interaction energy, one can see that the best
value of q~ is neither to be much smaller than, nor to be much larger than rl~,.
A natural choice then would be the average of these two situations, rl~ ,~ q ] ,
which is precisely the HSAB principle.
It is important to mention that using Eq. (53), and a qualitative explanation
of the equivalent to Eq. (55), Chattaraj, Lee and Parr [35] have arrived to the
same conclusion with respect to the HSAB principle.
On the other hand, they have also given an additional proof that is based on
Eq. (53) rewritten in the form of Eq. (43) with

AE A = - 89 - ~t~)2 rl~,
(n~, + ri~) 2 (56)

and

AE~ = - ~(~t~ - ~ , ) 2 n~
(1]~ "-{-T[~) 2 (57)

By assuming that for a given (I.t] - ~t~), and rl~, AE~ is minimized with respect
to rl~, one finds that q~, = q~. The minimization of AE~ with respect to rl~ with
rl~, fixed leads to the same result. Thus, if one assumes that AE~ and AE~
separately like to be as negative as possible, then Eqs. (56) and (57) lead to the
HSAB principle.
The same procedure may be used in connection with the contribution given
by Eq. (55). In this case,
AE. = AE A + AE.B (58)
where

AE A = _ 12 NAB
2 k s~
(St~ "]- S~)2 (59)

and

AE,n = --• 2 k s~ (60)


2 AB (S~ "1- S;) 2
o o o
If AE,A is minimized with respect to SA for a fixed SB, then one finds SA ----S~ that
is equivalent to rl~, = rl~. The minimization of AE~ with respect to s~ with
s~ fixed leads to the same result. Thus, it is found again that assuming that AE A
40 Jos6 L. Gtizquez

and AE~ separately like to be as negative as possible, then Eqs. (59) and (60) lead
to the HSAB principle.
The fact that the four terms, Eqs. (56)-(57) and Eqs. (59)-(60), reach their
minimum values at the same point, when rl~ = rl~, strongly suggests that there
will be a greater stabilization (AE~,t will be more negative) when the interacting
species have similar hardnesses, and this is precisely the HSAB principle.

7 Scaling Properties of Simple Functionals

The analysis presented in Sect. 5 will be strengthened in this section by showing


that under conditions of constant chemical potential the ground-state energy of
an atomic or molecular system is lowered when the hardness increases. This
statement may be seen as another form of the principle of maximum hardness.
The procedure is based on expressing the energy expression given in Eq. (1)
in terms of the chemical potential and the universal Hohenberg-Kohn func-
tional. This can be done multiplying Eq. (2) by p(r), integrating over r, and
substituting the resulting relation in Eq. (1). This procedure leads to Eq. (26),
which here is rewritten as [11, 28, 30]
E[p] = Nit + Q [ p ] (61)
where

Q [ p ] = F [ p ] - j'dr p(r)9-~-F, (62)


op~r)
Equation (61)establishes that under conditions of constant chemical potential,
the total energy will be directly proportional to Q[P]. Thus one needs to show
that the leading term of Q[p] is proportional to the global hardness. In order to
establish this result one makes use of the scaling properties of simple functionals
as follows.
For any well-behaved functional X [p] that obeys the scaling law
X [p,.(r)] = k" X [p(r)] (63)
where
p~(r) = X3 p(~r) (64)
one can show that [38]

8X (65)
X[p] = - n ~ dr p(r)r" V 89(r-----~

Following the arguments given by Kugler [39], one can express Eq. (65) in
several forms, that are derived through a set of equations that are obtained from
Hardness and Softnessin Density Functional Theory 41

repetitive functional differentiation of Eq. (65). In particular, one can eliminate


all the gradient terms to express X [ p l in the form of an infinite series,

5X
XEp] = S dr p(r) 8p(r)
1 52X
2 ~ dr dr' p(r) p(r') 8p(r') 8p(r)
83X
+ ~ Sj'~ dr dr' dr" p (r) p (r') p(r") 8p(r")8p(r')8p(r)

.... (66)
Note that this is an alternating series, independent of the value of n, which may
be hoped to be rapidly convergent. Equation (66) is a general representation of
a functional that satisfies the scaling law given by Eq. (63). Thus one may apply
it to F [p] under certain conditions, because not all the components of the exact
Hohenberg-Kohn functional obey this scaling law [30, 401. This can be seen
through the Kohn-Sham approach [10, 411, where the exact kinetic energy is
replaced by the non-interacting kinetic energy TsEPl, so that
F [ p ] = TsEPl + J[Pl + ExEPl + EcEPl (67)
where J[Pl is the coulombic interaction energy functional, ExEP] is the ex-
change energy functional, and Ec[p] is the correlation energy functional. In this
case one can show that [10, 30, 40]
Ts [p~.] = X2 Ts [ e l (68)
J[px] = L J [ p l (69)
and
Ex [P~.1 = ~ Ex [Pl (70)
However
Ec[0~] #- XEc[p] (71)
The correlation energy defined through Eq. (67) has a kinetic energy contribu-
tion, and does not behave as a purely potential energy term.
Returning to Eq. (66) one can see that this expression may be used for Ts[Pl,
J [Pl and Ex [p], but it can not be used for Ec [P]. However, if one considers the
exchange-only case, where Ec[Pl is neglected, then, because Eq. (66) is n-
independent, one can write
52F
QEP] ~ - 89SS drdr' p(r) p(r') 8p(r')8p(r) (72)

where third and higher order terms have been neglected, and F [ p l is approxi-
mated by Eq. (67) with Ec [P] = 0.
42 Jose L. G~izquez

The integral in Eq. (72) can be expressed in terms of the global hardness
through the use of Eq. (40), so that Eq. (61) can be written as

Eo ~ N~t -- 89N 211 (73)

This expression allows one to state that under conditions of constant


chemical potential the ground-state energy of a system decreases when the
hardness increases.
Starting from Eq. (73) one can also analyze energy changes associated to
different processes. In particular one can derive the expressions given in Sects. 5
and 6, although, formally in this case the correlation energy is neglected.

8 Concluding Remarks

Throughout this work we have studied the concepts of hardness and softness in
the framework of density functional theory. Thus it has been established that:
1) The fundamental equations to describe the change from one ground-state to
another provide a solid basis for defining hardness and softness functions
hierarchized as non-local, local and global quantities.
2) Through these definitions it has been established that under conditions of
constant chemical potential the interaction between two systems evolves
towards a state of maximum hardness, and that soft-soft and hard-hard
interactions are energetically favored.
3) Finally, it has been seen that to a good approximation the ground-state
energy of a system decreases when its hardness increases.
These general statements may prove very useful for analyzing reaction
mechanisms, and to understand the general behavior of a given molecule when it
interacts with different chemical species.
It is interesting to note that these principles may be used in the context of
wavefunction theory. In this case the global hardness may be approximated by
[42]

"11 ,~. CLUMO - - ~HOMO (74)


and therefore, the interaction mechanism may be analyzed in terms of the
maximum hardness principle, instead of the minimum energy principle. This
procedure may be particularly attractive in the case of semiempirical methods,
where the calculaltion of the frontier orbital energies is better than the calcu-
lation of the total energy. Some of the results that have been obtained through
this approach [6-8, 43] seem to indicate that the maximum hardness principle
may become a very useful concept in chemistry.
Hardness and Softness in Density Functional Theory 43

Acknowledgments. I would like to thank A. Vela, M. Galv~in, J. Robles, A.M. Martinez, and
F. M6ndez, for many valuable discussions. This work has been aided by a research grant from the
Consejo Nacional de Ciencia y Tecnologia.

9 References

1. Pearson RG (1973) Hard and soft acids and bases, Dowden, Hutchinson and Ross, Stroudsville,
PA
2. Sanderson RT (1976) Chemical bonds and bond energy, 2nd edn. Academic, New York
3. Fukui K (1973) Theory of orientation and stereoselection, Springer, Berlin Heidelberg New
York
4. Klopman G (ed) (1974) Chemical reactivity and reaction paths, Wiley, New York
5. Pearson RG (1987) J Chem Educ 64:561
6. Zhou Z, Parr RG, Garst JF (1988) Tetrahedron Lett 29:4843
7. Zhou, Z, Parr RG (1989) J Am Chem Soc 111:7371
8. Zhou Z, Parr RG (1990) J Am Chem Soc 112:5720
9. Parr RG, Chattaraj PK (1991) J Am Chem Soc 113:1854
10. Parr RG, Yang W (1989) Density functional theory of atoms and molecules, Oxford University
Press, New York
11. Parr RG, Donnelly RA, Levy M, Palke WE (1978) J Chem Phys 68:3801
12. Parr RG, Pearson RG (1983) J Am Chem Soc 105:7512
13. Pearson RG (1985) J Am Chem Soc 107:6801
14. Pearson RG (1988) Inorg Chem 27:734
15. Lee C, Yang W, Parr RG (1988) J Mol Struc (Theochem) 163:305
16. Mdndez F, Galv~.n M (1991) In: Labanowski JK, Andzelm JW (eds) Density functional methods
in chemistry. Springer, Berlin Heidelberg New York, p 387
17. M6ndez F, Galv~in M, Garritz A, Vela A, Gfizquez JL (1992) J Mol Struc (Theochem) in press.
18. Nalewajski RF, Korehowiec J, Zhou Z (1988) lnt J Quantum Chem $22:349
19. Nalewajski RF (1989) J Phys Chem 93:2658
20. Nalewajski RF, Korchowiec J (1989) J Mol Catal 54:324
21. Yang W, Mortier WJ (1986) J Am Chem Soc 108:5708
22. Mortier WJ, Ghosh SK, Shankar S (1986) J Am Chem Soc 108:4315
23. Berkowitz M, Parr RG (1988) J Chem Phys 88:2554
24. Pearson RG (1989) J Org Chem 54:1423
25. Hohenberg P, Kohn W (1964) Phys Rev 136:B864
26. Yang W, Parr RG (1985) Proc Natl Acad Sci USA 82:6723
27. Parr RG, Yang W (1984) J Am Chem Soc 106:4049
28. Berkowitz M, Ghosh SK, Parr RG (1985) J Am Chem Soc 107:6811
29. Ghosh SK (1990) Chem Phys Lett 172:77
30. Levy M, Perdew JP (1985) Phys Rev A 32:2010
31. Deb BM (ed) (1981) The Force Concept in Chemistry, Van Nostrand Reinhold, New York
32. G~tzquez JL, Galvfin M, Vela A (1990) J Mol Struc (Theochem) 210:29
33. Yang W, Parr RG, Pucci R (1984) J Chem Phys 81:2862
34. Pearson RG (1963) J Am Chem Soe 85:3533
35. Chattaraj PK, Lee H, Parr RG (1991) J Am Chcm Soe 113:1855
36. Yang W, Lee C, Ghosh SK (1985) J Phys Chem 89:5412
37. Fuentealba P, Parr RG (1991) J Chem Phys 94:5559
38. Ghosh SK, Parr RG (1985) J Chem Phys 82:3307
39. Kugler AA (1990) Phys Rev A 41:3489
40. Levy M, Yang W, Parr RG (1985) J Chem Phys 83:2334
41. Kohn W, Sham l_J (1965) Phys Rev 140:A1133
42. Pearson RG (1986) Proc Natl Acad Sei USA 83:8440
43. Martinez AM, Vela A, Robles J (private communication)

You might also like