Documente Academic
Documente Profesional
Documente Cultură
=
_
Z
i
Z
j
e
(r)
r/
if r ,
if r < ,
(1)
where U is the inter-particle potential energy, Z
i
and Z
j
are the
charges on the particles i and j, respectively, r is the distance
between particles i and j, is the inverse screening length,
is the diameter of each particle and is energy when two
particles with charges equal to unity are in contact with each
other.
B. Phase equilibrium
The conditions for phase equilibria between two phases
of a binary mixture are the equalities of temperatures, pres-
sures, and the chemical potentials of both components in the
two phases. In our study, the uid phase and the substitution-
ally disordered solid phases are simulated in the semigrand
ensemble keeping the total number of particles xed. Hence,
we rewrite the phase equilibrium criteria between two phases,
and , containing components A and B as follows.
T
()
= T
()
, (2a)
P
()
= P
()
, (2b)
()
=
()
, (2c)
()
B
=
()
B
. (2d)
In the above equations, T is the temperature, P is the pres-
sure,
i
is the chemical potential of component i, and
=
A
B
. The above equations are valid for phase equilib-
rium between uid phase and the substitutionally disordered
solid phase. If the solid phase is a substitutionally ordered and
has a stoichiometric ratio of 1 : n between components A and
B, then the composition of the solid phase is invariant and
hence Eqs. (2c) and (2d) will be replaced by
28
()
A
+n
()
B
= (1 +n)g
()
, (3)
where is the substitutionally ordered solid phase and g is
the molar Gibbs free energy of the solid.
C. Equations of state
The equations of state for the uid and solid phases are
obtained using a combination of Monte Carlo molecular sim-
ulations and the method of thermodynamic integration. Since
both the uid and the solid phases have high density, we nd
it most convenient to carry out the simulations in the isother-
mal isobaric semigrand ensemble. In this ensemble, the simu-
lations are carried out keeping temperature (T), pressure (P),
chemical potential difference (), and total number of par-
ticles (N) constant. Integrating the Gibbs-Duhem equation at
constant temperature, the form of the equation of state can be
written as
B
(T, P, ) =
B
(T, P
0
,
0
)
__
0
x
A
d
_
T,P
0
+
__
P
P
0
V dP
_
T,
, (4)
where P
0
is the pressure of the reference state,
0
is the
chemical potential difference between the components in the
Downloaded 16 Apr 2012 to 203.200.35.12. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
134506-3 Pavaskar, Sharma, and Punnathanam J. Chem. Phys. 136, 134506 (2012)
reference state, V is the specic volume, and x
A
is the mole
fraction of component A. The values of the integrands, V
and x
A
are obtained from the simulations. The reference state
used in Eq. (4) is pure B phase. This results in the value
of
0
being equal to . This causes practical problems
in straightforward numerical evaluation of the rst integral
on RHS of Eq. (4). Equation (4) is hence rewritten as (see
Appendix)
B
(T, P, ) =
B
(T, P
0
,
0
) +ln(1 x
A
)
x
A
ln
_
B
_
+
__
x
A
0
ln
_
B
_
dx
A
_
T,P
0
+
__
P
P
0
V dP
_
T,
, (5)
where
i
is the activity coefcient of component i, = 1/k
B
T
and k
B
is the Boltzmanns constant. The ratio (
A
/
B
) is eval-
uated from the semigrand simulations as
ln
_
B
_
=
_
(p)
A
(p)
B
_
ln
_
x
A
1 x
A
_
, (6)
where
(p)
A
and
(p)
B
are the chemical potentials of A and B in
pure phases, respectively, at T and P
0
, and k
B
is the Boltzmann
constant. In order to evaluate the integral in Eq. (5), the values
of ln
_
B
_
obtained from simulations is tted by a polynomial
of x
A
under the constraint
_
1
0
ln
_
B
_
dx
A
= 0. (7)
For the case of substitutionally ordered solids, the sim-
ulations are performed in the isothermal-isobaric ensemble.
The form of the equation of state is
g(T, P) = g(T, P
0
) +
__
P
P
0
V dP
_
T
. (8)
The values of the molar volume, V, is obtained from the
isothermal-isobaric simulations.
To complete the determination of the equations of state,
it is necessary to compute the chemical potentials and free en-
ergies at the reference state. For the uid phases, the chemical
potentials of the components in their pure phases were com-
puted using the method of Hamiltonian integration.
32
The ref-
erence uid used in this method is the hard sphere uid. For
solid phases, the reference state properties are computed us-
ing the Frenkel-Ladd
33, 34
method. The Frenkel-Ladd method
is widely used to compute the free-energies of solid phases.
D. Crystal structure of solid phases
The method described in Sec. II C computes the phase
equilibrium between the uid phase and a solid phase with
known crystal structure. Hence, computation of the solid-uid
phase diagram requires a priori knowledge of all possible
crystalline structures. We have simulated a wide variety of
solid phases, both substitutionally ordered and substitution-
ally disordered, in this study. Substitutionally ordered phases
include the CuAu phase, Cu3Au phase, CsCl phase, NaCl
phase, and the face-centered tetragonal structures.
25, 35
The
substitutionally disordered phases include FCC and body-
centered cubic structures. We have not simulated the substi-
tutionally disordered hexagonal-close-packed (HCP) phase.
Earlier studies on oppositely charged colloids
25
have reported
the structure of the substitutionally disordered phase to be
FCC. Moreover, in the case of hard spheres, it has been found
that although FCC is more stable than HCP, the free energy
difference between the two is less than 10
3
k
B
T. Hence, for
the sake of convenience, we have assumed that the FCC phase
is more stable than HCP phase.
E. Simulation details
The simulations of both uid and substitutionally disor-
dered solid phases were carried out in the isothermal-isobaric
semigrand ensemble (constant N, P, T, ) and those of sub-
stitutionally ordered solid phases were done in the isothermal-
isobaric ensemble (constant N, P, T). In the isothermal iso-
baric ensemble, the Monte Carlo moves used were particle
displacements and isotropic volume changes. In the semi-
grand ensemble, identity change moves were used in addition
to particle displacements and volume moves. Each simulation
consisted of at least 100 000 equilibration cycles and 500 000
production cycles. Each cycle consists of N particle moves
(i.e., translation or identity change) and one volume move.
The averages reported are over ve independent simulations.
III. RESULTS AND DISCUSSION
The solid-uid phase behavior of oppositely charged col-
loidal suspensions were studied by computing the pressure-
composition phase diagrams. The computations were per-
formed at a reduced temperature T*( = k
B
T/) equal to 1.0
and 2.0. The solid-uid phase equilibrium of our system de-
pends on factors such as ratio of the charges on the particles,
Z
B
/Z
A
, and reduced screening length, . In order to study the
effect of these factors, the phase diagrams were computed for
various values of Z
B
/Z
A
and .
We rst discuss the phase diagrams at T* = 1.0.
Figures 15 show the pressure-composition phase diagrams
for the various values of |Z
B
/Z
A
| and . The numerical val-
ues for some of coexistence points are given in Tables IIV.
For the range of pressures studied, only two kinds of solid
phases namely, the substitutionally disordered FCC phase and
the substitutionally ordered CsCl phase have been observed.
Since the particles are oppositely charged, as expected, the
freezing pressure of the mixture is lower than that of the pure
uid. In other words, it is easier for the mixture to form a solid
compared to a pure uid.
Figure 1 shows the phase diagram for = 3.0 and
|Z
B
/Z
A
| = 1.0. Two kinds of solid phases are observed de-
pending upon the external conditions. These are the substi-
tutionally ordered CsCl phase and the substitutionally disor-
dered FCC phase. The formation of the CsCl phase from the
suspension is favored when the mole fractions of the two com-
ponents are nearly equal. When one of the components is pre-
dominant over the other, then the substitutionally disordered
Downloaded 16 Apr 2012 to 203.200.35.12. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
134506-4 Pavaskar, Sharma, and Punnathanam J. Chem. Phys. 136, 134506 (2012)
11
12
13
14
15
16
17
18
19
0 0.2 0.4 0.6 0.8 1
P
3
x
A
FCC
uid
CsCl
FIG. 1. P-x phase diagram for a binary mixture of equal-sized and oppositely
charged Yukawa particles at T* =1.0. The values of and |Z
B
/Z
A
| are equal
to 3.0 and 1.0, respectively. The red symbols show equilibrium between the
uid phase and the substitutionally disordered FCC solid phase. The green
symbols show equilibrium between the uid phase and the substitutionally
ordered CsCl solid phase. The blue symbols show equilibrium between the
substitutionally disordered FCC solid and the substitutionally ordered CsCl
solid. The dotted lines serve as a guide to the eye. The size of errorbars rep-
resent 95% condence limit.
FCC phase is favored. Even when the CsCl phase is formed,
it is thermodynamically stable only for a moderate range of
pressures. As the pressure is increased, the CsCl phase under-
goes phase transition to the FCC phase. Figure 2 shows the
phase diagram for = 3.0 and |Z
B
/Z
A
| = 0.8. The phase di-
agram is spindle shaped with an azeotrope and the only solid
phase that is formed is the substitutionally disordered FCC
phase.
At higher values of charge screening, i.e., at = 6.0,
the corresponding phase diagrams for values of |Z
B
/Z
A
| equal
to 1.0, 0.8, and 0.6 are shown in Figs. 35, respectively. In
this case, the suspension has the ability to form the CsCl solid
for values of |Z
B
/Z
A
| equal to 1.0 and 0.8. This shows that an
11
12
13
14
15
16
17
18
19
0 0.2 0.4 0.6 0.8 1
P
3
x
A
FCC
uid
FIG. 2. P-x phase diagram for a binary mixture of equal-sized and oppositely
charged Yukawa particles at T* =1.0. The values of and |Z
B
/Z
A
| are equal
to 3.0 and 0.8, respectively. The size of errorbars represent 95% condence
limit. The explanations for lines and symbols are same as Fig. 1.
8
10
12
14
16
18
20
22
0 0.2 0.4 0.6 0.8 1
P
3
x
A
FCC
uid
CsCl
FIG. 3. P-x phase diagram for a binary mixture of equal-sized and oppositely
charged Yukawa particles at T* =1.0. The values of and |Z
B
/Z
A
| are equal
to 6.0 and 1.0, respectively. The size of errorbars represent 95% condence
limit. The explanations for lines and symbols are same as Fig. 1.
8
10
12
14
16
18
20
22
0 0.2 0.4 0.6 0.8
P
3
x
A
FCC
uid
CsCl
FIG. 4. P-x phase diagram for a binary mixture of equal-sized and oppositely
charged Yukawa particles at T* =1.0. The values of and |Z
B
/Z
A
| are equal
to 6.0 and 0.8, respectively. The size of errorbars represent 95% condence
limit. The explanations for lines and symbols are same as Fig. 1.
8
10
12
14
16
18
20
22
0 0.2 0.4 0.6 0.8 1
P
3
x
A
FCC
uid
FIG. 5. P-x phase diagram for a binary mixture of equal-sized and oppositely
charged Yukawa particles at T* =1.0. The values of and |Z
B
/Z
A
| are equal
to 6.0 and 0.6, respectively. The size of errorbars represent 95% condence
limit. The explanations for lines and symbols are same as Fig. 1.
Downloaded 16 Apr 2012 to 203.200.35.12. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
134506-5 Pavaskar, Sharma, and Punnathanam J. Chem. Phys. 136, 134506 (2012)
TABLE I. Phase equilibria between uid phase and FCC phase for pure
components at various values of and |Z
B
|.
|Z
B
| P
3
/
_
3
_
uid
_
3
_
FCC
3.0 1.0 18.00 27.23 0.951 1.021
3.0 0.8 15.87 23.41 0.950 1.030
6.0 1.0 16.42 23.93 0.941 1.008
6.0 0.8 14.97 21.41 0.946 1.023
6.0 0.6 13.72 19.33 0.947 1.034
increase in the value of from 3.0 to 6.0 favors the forma-
tion of substitutionally ordered CsCl phase. However, from
Fig. 4 we see that the range of compositions for which the
uid phase is in equilibrium with the CsCl phase is drastically
reduced when the asymmetry in the magnitude of the charges
between oppositely charged particles reached just 20%. When
|Z
B
/Z
A
| = 0.6 the only thermodynamically stable solid phase
is the disordered FCC phase and the phase diagram is spindle
shaped with an azeotrope. Also as before whenever the CsCl
phase is formed, it undergoes a phase transition to the sub-
stitutionally disordered FCC phase with increase in pressure.
The CsCl phase has a body centered cubic structure and as the
pressure is increased, its density approaches the closed packed
density,
3
of 1.299. At this density, the FCC phase has
higher free volume and hence a higher entropy. This leads to
a phase transition from the CsCl phase to the substitutionally
disordered FCC phase. Earlier studies
25
had reported a phase
transition from substitutionally ordered body-centered CsCl
phase to substitutionally ordered face-centered CuAu phase.
However, our simulations always showed the substitutionally
disordered FCC phase to be more stable than the substitution-
ally ordered CuAu phase. The study by Hynninen et al.
25
also
shows that at still higher pressures, a substitutionally ordered
face-centered tetragonal phase should appear. Since the main
focus was the formation of substitutionally ordered structures
fromthe uid phase, solid-solid phase transitions at these high
pressures were not studied here.
As the value of is increased, the suspension begins to
resemble one made up of hard spheres. At the sus-
pension is similar to a hard sphere system. We know that for
such systems, the only stable solid phase is the FCC phase.
33
This in conjunction with the phase diagrams computed above
seems to indicate that there exists a range of values for ,
i.e., charge screening, that favors the formation of substitu-
tionally ordered superlattices when the particles are equally
sized.
Phase diagrams for T* = 2.0 were also computed for
various values of |Z
B
/Z
A
| and /. Our calculations (results
not included) showed that at this temperature, only the sub-
stitutionally disordered FCC phase in equilibrium with the
TABLE II. Phase equilibria between uid phase at 50% composition and
CsCl phase at various values of and |Z
B
/Z
A
|.
|Z
B
/Z
A
| P
3
/
A
B
_
3
_
uid
_
3
_
CsCl
3.0 1.0 11.33 14.60 14.60 0.945 1.030
6.0 1.0 9.93 13.11 13.11 0.921 1.031
6.0 0.8 11.53 15.60 14.38 0.948 1.046
TABLE III. Phase equilibria between FCC phase at 50% composition and
CsCl phase at various values of and |Z
B
/Z
A
|.
|Z
B
/Z
A
| P
3
/
A
B
_
3
_
FCC
_
3
_
CsCl
3.0 1.0 13.2 16.4 16.4 1.091 1.059
6.0 1.0 21.5 23.6 23.6 1.209 1.152
6.0 0.8 15.2 19.0 17.8 1.122 1.095
uid phase and the phase diagram is spindle shaped with an
azeotrope. This is because the contribution to the free energy
due to entropy increases with rise in temperature. Hence the
substitutionally disordered FCC phase is thermodynamically
more stable than the CsCl phase. At lower temperatures, the
phase diagrams become more complex due to the appearance
of new phases such as Cu3Au.
26
We are currently carrying
detailed investigations on phase diagrams where the thermal
energy is lower than the contact energy of the particles and
the results will be published later.
An analysis of the phase diagrams (Figs. 15) leads to
two major observations. The rst observation is that the CsCl
phase is favored only when |Z
B
/Z
A
| is close to the stoichiomet-
ric ratio of one. This can be explained in terms of the inter-
play between the entropy and the energy of the solid phases.
The substitutional order of the CsCl phase naturally leads to a
lower entropy compared to the FCC phase. In the CsCl phase,
every particle is surrounded by eight oppositely charged par-
ticles. In order for the CsCl phase to be more stable, this has
to be compensated by lowering of the internal energy due to
interaction among oppositely charged particles. However, as
the charge asymmetry increases, i.e., |Z
B
/Z
A
| 0, the inter-
nal energy of the solid phase rises due to decreased interaction
among oppositely charged particles. Hence, we see the disap-
pearance of CsCl phase from the phase diagrams for lower
values of |Z
B
/Z
A
|. This has important implications with re-
gards to synthesis of substitutionally ordered solids. Although
unlike ionic systems, the stoichiometry of substitutionally or-
dered superlattices do not dictate the charge ratio, these results
show that the allowed deviation of the charges from the sto-
ichiometric ratio may be limited especially when the contact
energy is of the order of k
B
T.
The second observation is that the formation of CsCl
phase is more favorable at some intermediate range of charge
screening. As mentioned earlier, the CsCl phase is formed if
the lowering of the internal energy caused due to attraction
between oppositely charged particles compensates the lower
entropy due to substitutional order. If the screening is very
high, then the interaction among particles is very low and
TABLE IV. Triple points between the uid, FCC and CsCl phases at various
values of and |Z
B
/Z
A
|.
|Z
B
/Z
A
| P
3
/ x
uid
A
x
FCC
A
3.0 1.0 11.77 0.461 0.465
3.0 1.0 11.77 0.539 0.535
6.0 1.0 12.35 0.376 0.391
6.0 1.0 12.35 0.624 0.609
6.0 0.8 11.99 0.444 0.444
6.0 0.8 12.25 0.569 0.553
Downloaded 16 Apr 2012 to 203.200.35.12. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
134506-6 Pavaskar, Sharma, and Punnathanam J. Chem. Phys. 136, 134506 (2012)
internal energy becomes negligible. This favors the FCC
phase which has higher entropy. At very low screening,
each particle not only interacts with its immediate oppositely
charged neighbors but also with like charged particles beyond
the nearest neighbors. As a result the overall internal energy
is not sufciently lowered to compensate for loss of entropy
due to substitutional order. Thus the CsCl phase is favored at
some intermediate screening. A similar explanation has been
proposed for the formation of low density crystalline aggre-
gates from binary nanoparticles suspensions in the work of
Kalsin et al.
16
This observation is limited to equal-sized par-
ticles. Particle size ratio, which we have not studied here, also
plays a strong role in the formation of substitutionally ordered
superlattices.
28
IV. CONCLUSIONS
In this study, we have examined the effect of electrostatic
interactions among colloidal particles on the structure of su-
perlattices. Towards this end, pressure-composition phase dia-
grams of mixtures of equal sized and oppositely charged col-
loidal particles interacting via the hard core Yukawa poten-
tial have been computed. These phase diagrams show the ef-
fect of composition, charge asymmetry, and charge screening
towards the structure of superlattices formed by oppositely
charged particles. The studies were done under conditions
where the thermal energy, k
B
T, is equal or greater than in mag-
nitude to the contact energy between the particles and the De-
bye screening length is smaller than the size of the particles.
Two signicant conclusions emerge from our study. First, we
nd that the ability of the suspensions to form substitution-
ally ordered solids is sharply diminished when the ratio of
the magnitude of the charges differs from unity, which is the
stoichiometric ratio of the solid. Hence even if the stoichio-
metric ratio of the substitutionally ordered solid does not dic-
tate the charge ratio, the allowed deviation is small. The other
signicant conclusion is that the formation of substitutionally
ordered superlattices from equal sized particles is favored at
moderate levels of charge screening. At low and high values
of charge screening the formation of the substitutionally dis-
ordered FCC phase is favored.
ACKNOWLEDGMENTS
The nancial support for this work has been provided by
grants from the Department of Science and Technology, Gov-
ernment of India.
APPENDIX: EQUATION OF STATE FOR FLUIDS AND
SUBSTITUTIONALLY DISORDERED SOLIDS
Consider a binary mixture containing components A and
B. The chemical potentials of both these components can be
written as
A
(T, P, x
A
) =
(p)
A
(T, P) +ln[x
A
A
], (A1)
B
(T, P, x
A
) =
(p)
B
(T, P) +ln[(1 x
A
)
B
]. (A2)
From the above equations, the expression for ( =
A
B
) can be written as
=
_
(p)
A
(p)
B
_
+ln
_
x
A
1 x
A
B
_
. (A3)
Substituting the above expression into the rst integral in
Eq. (4) and integrating by parts, we get
__
0
x
A
d
_
T,P
0
= ln(1 x
A
) x
A
ln
_
B
_
+
__
x
A
0
ln
_
B
_
dx
A
_
T,P
0
.
(A4)
1
P. N. Pusey and W. van Megen, Nature (London) 320, 340 (1986).
2
B. J. Alder and T. E. Wainwright, J. Chem. Phys. 27, 1208 (1957).
3
W. G. Hoover and F. H. Ree, J. Chem. Phys. 49, 3609 (1968).
4
P. Bartlett and R. H. Ottewill, J. Chem. Phys. 93, 1299 (1990).
5
P. Bartlett, R. H. Ottewill, and P. N. Pusey, Phys. Rev. Lett. 68, 3801
(1992).
6
C. B. Murray, C. R. Kagan, and M. G. Bawendi, Science 270, 1335
(1995).
7
M. Brust, R. Etchenique, E. J. Calvo, and G. J. Gordillo, Chem. Commun.
1996(16), 1949.
8
C. J. Kiely, J. Fink, M. Brust, D. Bethell, and D. J. Schiffrin, Nature
(London) 396, 444 (1998).
9
Z. L. Wang, Adv. Mater. 10, 13 (1998).
10
N. Hunt, R. Jardine, and P. Bartlett, Phys. Rev. E 62, 900 (2000).
11
A. B. Schoeld, Phys. Rev. E. 64, 051403 (2001).
12
F. X. Redl, K. S. Cho, C. B. Murray, and S. OBrien, Nature (London) 423,
968 (2003).
13
P. Bartlett and A. I. Campbell, Phys. Rev. Lett. 95, 128302 (2005).
14
M. E. Leunissen, C. G. Christova, A. P. Hynninen, C. P. Royall, A. I.
Campbell, A. Imhof, M. Dijkstra, R. van Roij, and A. van Blaaderen,
Nature (London) 437, 235 (2005).
15
E. V. Shevchenko, D. V. Talapin, N. A. Kotov, S. OBrien, and C. B.
Murray, Nature (London) 439, 55 (2006).
16
A. M. Kalsin, M. Fialkowski, M. Paszewski, S. K. Smoukov, K. J.
M. Bishop, and B. A. Grzybowski, Science 312, 420 (2006).
17
H. Zhang, E. W. Edwards, D. Wang, and H. Mohwald, Phys. Chem. Chem.
Phys. 8, 3288 (2006).
18
A. P. Hynninen, C. G. Christova, R. van Roij, A. van Blaaderen, and M.
Dijkstra, Phys. Rev. Lett. 96, 138308 (2006).
19
A. Yethiraj, Soft Matter 3, 1099 (2007).
20
M. P. Pileni, Acc. Chem. Res. 40, 685 (2007).
21
Z. Chen, J. Moore, G. Radtke, H. Sirringhaus, and S. OBrien, J. Am.
Chem. Soc. 129, 15702 (2007).
22
K. Overgaag, W. Evers, B. de Nijs, R. Koole, J. Meeldijk, and D.
Vanmaekelbergh, J. Am. Chem. Soc. 130, 7833 (2008).
23
E. C. M. Vermolen, A. Kuijk, L. C. Filion, M. Hermes, J. H. J. Thijssen,
M. Dijkstra, and A. van Blaaderen, Proc. Natl. Acad. Sci. U.S.A. 106,
16063 (2009).
24
D. Vanmaekelbergh, Nanotoday 6, 419 (2011).
25
A. P. Hynninen, M. E. Leunissen, A. van Blaaderen, and M. Dijkstra, Phys.
Rev. Lett. 96, 018303 (2006).
26
M. Bier, R. van Roij, and M. Dijkstra, J. Chem. Phys. 133, 124501 (2010).
27
L. Filon and M. Dijkstra, Phys. Rev. E 79, 046714 (2009).
28
X. Cottin and P. A. Monson, J. Chem. Phys. 102, 3354 (1995).
29
B. Derjaguin and L. Landau, Acta Physicochim. URSS 14, 633 (1941).
30
E. J. W. Verwey and J. T. G. Overbeek, Theory of the Stability of Lyotropic
Colloids (Elsevier, 1948).
31
A. P. Hynninen and M. Dijkstra, Phys. Rev. E 68, 021407 (2003).
32
C. Vega, E. Sanz, J. L. F. Abascal, and E. G. Noya, J. Phys.: Condens.
Matter 20, 153101 (2008).
33
D. Frenkel and A. J. C. Ladd, J. Chem. Phys. 81, 3188 (1984).
34
J. M. Polson, E. Trizac, S. Pronk, and D. Frenkel, J. Chem. Phys. 112, 5339
(2000).
35
F. Bresme, C. Vega, and J. L. F. Abascal, Phys. Rev. Lett. 85, 3217 (2000).
Downloaded 16 Apr 2012 to 203.200.35.12. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions