• Aucun résultat trouvé

Interchain coupling in polyacetylene

N/A
N/A
Protected

Academic year: 2021

Partager "Interchain coupling in polyacetylene"

Copied!
8
0
0

Texte intégral

(1)

HAL Id: jpa-00209951

https://hal.archives-ouvertes.fr/jpa-00209951

Submitted on 1 Jan 1985

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Interchain coupling in polyacetylene

P.L. Danielsen, R.C. Ball

To cite this version:

P.L. Danielsen, R.C. Ball. Interchain coupling in polyacetylene. Journal de Physique, 1985, 46 (2),

pp.131-137. �10.1051/jphys:01985004602013100�. �jpa-00209951�

(2)

131

LE JOURNAL DE PHYSIQUE

Interchain coupling in polyacetylene

P. L. Danielsen and R. C. Ball

Cavendish Laboratory, Madingley Road, Cambridge CB3 OHE, U.K.

(Reçu le 27 juin 1984, accepté le 17 octobre 1984)

Résumé.

2014

Nous considérons le couplage entre deux chaînes de polyacetylène, (i) pour les chaînes sans solitons et

(ii) pour les chaînes chacune avec un soliton neutre. Dans le premier cas nous trouvons une correction nouvelle pour la grandeur et pour la position de la bande interdite. Dans le cas avec solitons, nous trouvons une interaction

entre leurs niveaux au centre de la bande interdite. Les différences entre le cas de dimérisation alignée à ± ~ et le

cas de dimérisation décalée sont démontrées. La séparation entre les niveaux est 0394E = t~ l/03BE sinh l/03BE, dans le cas aligné, et ~ 0(e-kf03BE) dans le cas décalé (où l est la séparation entre les solitons, kf le vecteur d’onde de Fermi, 03BE la

taille du soliton, et [1] kf 03BE ~ 10). Cette interaction contribue à la localisation tridimensionnelle des solitons,

ce qui renforce l’argument contre le modèle solitonique de conduction dans le polyacétylène. Ainsi nous prévoyons

aussi l’existence d’une densité critique de solitons 03C1c ~ 0,05; où l’ordre local favorisé entre chaînes devient aligné

au lieu de décalé.

Abstract.

2014

The interchain coupling between two polyacetylene chains is considered, both in the absence of

solitons, and with one neutral soliton on each chain. A new correction to the size and position of the band gap is found in the former case, and when solitons are present it is shown that their mid-gap levels interact. In this case

the differences between the matched (same dimerization pattern at ± ~) and the mismatched cases (opposite pattern at ± ~) are clearly illustrated. The mid-gap level splitting is found to be 0394E

=

t~ l/03BE sinh l/03BE, in the

matched case, and ~ 0(e-kf03BE) in the mismatched case (where l is the soliton separation, kf the Fermi wavevector, 03BE the soliton width, and [1] kf 03BE ~ 10). This interaction increases the strength of the 3 D pinning of solitons, there-

fore strengthening the case against the solitonic model for conduction in polyacetylene. It also predicts the existence of a critical density 03C1c ~ 0.05 of solitons, at which the preferred local ordering between chains switches from mismatched to matched dimerization.

J. Physique 46 ( 1985) 131-137 FTVRIER 1985,

Classification

Physics Abstracts

72 . 80L

1. Introductioa

There has been considerable theoretical interest in

polyacetylene, (CH)x, in recent years, it being the simplest of a now growing class of conducting conju- gated polymers. The discrete soliton model of Su, Schrieffer, and Heeger [1] (SSH) and of Rice [2] (and

the extension to the continuum limit by Takayama,

Lin Liu, and Maki [3] (TLM)) is attractive, both for

its simplicity and for its apparent success in explaining

much of the magnetic, electric, and optical data for trans-(CH)x, though there are many discrepancies

with experiment [4]. It is possible that these could be

explained by including quantum fluctuations, electron correlations, lattice deformation terms, and three dimensional ordering. The relative size of the intra- and inter-chain hopping terms, to and t.1’ estimated

to be - 2.5 eV and - 0.1 eV respectively [1, 5],

suggest that conduction in three dimensions should be relatively easy. This, together with X-ray scattering

results which suggest that the dimerisation pattern shows 3 D order [6], has drawn much of the recent interest to 3 D effects.

There have been several publications on the subject [7, 8] which treat the model of two parallel (CH)x

chains where coupling is due to interchain hopping

of the n-electrons. They show that this mechanism favours mismatched over matched ordering of the

chains. Also, the confinement energy of two solitons

on mismatched (opposite dimerisation pattern at + oo ) chains is found to be

where a is the lattice spacing, I the soliton separation, and ç the soliton width. However, several important

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01985004602013100

(3)

132

points were overlooked, including the interaction of midgap states between matched chains, and the subsequent marked differences between the matched and mismatched cases.

The object of this paper is to both clarify and to

extend some of these previous results. Section 2 treats the case of two pristine (CH)x chains and summarizes the derivation of the preference of mismatched order-

ing. Some interesting new effects on the dependence

of both the size and position of the band gap on the

ordering of the chains are then presented Section 3

treats the confinement energy of two neutral solitons

on different chains finding some new and important

terms. We illustrate the differences between the two

possible orderings, and calculate the splitting of the midgap levels in the matched case. The confinement energy of the two solitons is then found-showing that

the matched and mismatched cases differ considerably, leading to a critical concentration Pc of solitons, above which matched ordering is preferred. Consequences

of these results are then discussed

2. Interchain coupling between 2 pristine (CH)x chains.

The model considered [7] is that of two parallel (CH)x

chains (j

=

1, 2) described individually by the SSH

Hamiltonian [1] ]

(where the spin suffixes have been suppressed for simplicity), and coupled by an interchain hopping

term

We then obtain the simple ladder hopping problems

shown in figure 1. In the notation of SSH, to is the hopping integral for the undimerized chain, and tl

=

2 au where a is the electron-phonon coupling

constant and u is the displacement of the carbon atoms

Fig. 1.

-

Ladder models for : a) matched ordering, and b)

mismatched ordering.

from their undimerized positions. It is assumed

throughout that tl tl.

In the case of a single chain (1) we had, in the basis

(01, OC),

In the case of the ladder problems, if we use the basis (03C8v1, 03C8c1, 03C8v2, 03C8c2), the Hamiltonian takes the form :

where 6 = + /

-

1 for matched/mismatched ordering.

Clearly, the matched case will simply lead to bonding

and antibonding states, with energies

(where Ek

=

± (82 + j2), /2 is the single chain energy)

as shown in figure 2a. Thus, as both the bonding

and antibonding states are fully filled, the net change

in energy, AE matched

=

0.

Fig. 2.

-

Band structure for : a) matched case, and b) mismatched case.

(4)

The splitting of the energy levels in the mismatched

case is easily obtained by diagonalizing (5) with

6

= -

1, giving :

which behave as shown in figure 2b. In this (mismat- ched) case, the presence of two interacting chains

will affect the size of tl, but we assume that this is

negligible compared to the energy directly associated

with the interchain hopping.

So, neglecting the shift in tj, the difference in energy from the two individual chains is given by :

giving, after integration over the first Brillouin zone,

(in agreement with the results obtained by Baeriswyl

and Maki [7] using the substitution

and diagonalizing using a Bogoliubov transforma-

tion). Thus mismatched ordering is preferred to

matched ordering by an amount W ~ 1.3 x l0-3 eV

or 15 K per atom. This is in agreement with results found for coupling due to coulombic forces [9, 10]

where the mismatched ordering was also found to be

preferred. Even allowing for the uncertainty in the

value of t.1’ which has been reported to be as low as

25 meV [11] (giving W - 1 K), coupling due to inter-

chain hopping will be the stronger effect.

The effect of the interchain coupling on the band

structure is shown in figure 2. This illustrates the two main differences between the two types of ordering : Firstly, the size of the band gap depends on the ordering :

and as t1 ~ 0.35 eV [1], Eg - 1.2 eV and 1.4 eV respectively, the band gap being larger in the preferred

mismatched case.

Secondly, the position of the band gap changes,

from being at the Fermi wavevector kf to a position at kf ± q, where

3. Confinement energy of two solitons on different chains.

3.1 WAVEFUNCTIONS FOR THE MATCHED AND MIS- MATCHED CASES.

-

The problem considered is that of two parallel (CH)x chains, both containing a

soliton defect. There are two (non equivalent) cases

of interest, those of the matched and mismatched

cases shown in figure 3, where 0 denotes a soliton,

and A and B are the two different (though degenerate) phases of (CH)x.

The wavefunctions of a single chain, can be found

using the continuum approximation [3]. For a chain

of length L, with a soliton centred at the origin they

are given by [7] (in spinor representation where

for the bound midgap state, and

for the conduction band (upper sign) and the valence

band (lower sign) electrons. Here

and

and the wavevectors for the conduction and valence

Fig. 3.

-

Soliton-soliton confinement for : a) matched

case, and b) mismatched case.

(5)

134

bands satisfy the boundary conditions

respectively, with Ok

=

tan-’ kvfldo, where n, m are integers.

What happens when a soliton moves or the dime- rization pattern changes ?

When a soliton moves the underlying dimerization pattern does not change; and if one moves the dimeri- zation origin by any factor of 2a, there should be no

change in the physical results. Thus, there are two

distinct fixed positions in the problem, one which

locates the centre of the soliton, and the other which fixes the dimerization pattern at ± oo.

Hence, the wavefunctions should read exactly as

in equations (13) except the x’s should be replaced by (x - I) and equation (12) should now read

where I is the distance of the soliton from the origin,

and where d defines the dimerization pattern at

± oo ; if d = 0 or a multiple of 2a the dimerization pattern is unchanged, and if d is an odd multiple of a,

then the dimerization pattern is reversed.

Therefore, for a change in dimerization pattern (d = a), we have for the reverse chain (see Fig. 3)

This corresponds to the substitutions made by Baeriswyl and Maki [7], where d Z was replaced by

-

d 2 and t/J 2 by Q3 ql2l leaving the individual Hamil- tonians invariant. The physical significance of this symmetry is easy to see : changing the sign of d 2 has

the effect of reversing the dimerization pattern, which has, as shown above, the effect of changing the wave-

function to i 63 t/J 2.

3.2 SOLITON CONFINEMENT ENERGY. - It is argued

here that the interaction of the two (CH)x chains will be dominated by two effects :

Firstly, when the solitons are close, the interaction of the well localized midgap states (this will also depend, in general, on the soliton occupation numbers, though they are assumed to be neutral) ; and, secondly,

when the solitons are far appart, the effect of either

creating or destroying lengths of the favoured AB

phase. In the matched case the solitons will be encou-

raged to separate by the formation of the energetically preferred AB region, whereas in the mismatched case

they will be confined by this effect.

Using degenerate perturbation theory the splitting

of the midgap states is determined by

i.e.

In the matched case, for a soliton at x

=

0 on chain 1, and at x

=

I on chain 2,

so that

giving

However, in the mismatched case the wavefunctions

are now (for the same soliton positions as before)

so that

where kf ç ~ 10. Thus, the midgap splitting is of the

order of ti- for small soliton separations in the matched

case (see Fig. 4a), and is a negligible effect in the

Fig. 4.

-

a. Different competing effects in AEconf f in the matched case. b. AEconf for both matched and mismatched

cases (zero of energy in each case is arbitrary).

(6)

mismatched case, in agreement with the exact results obtained by Baeriswyl and Maki. As [5] t.1 ~ 0.05 to the effect in the matched case is expected to be signi-

ficant.

The details of the treatment of the mismatched case are given in the appendix, the results being (see also Fig. 4b)

which is, as expected, asymptotic to

at large ( ± ) distances, as predicted by simply consider- ing the effect of the annihilation of AB phase.

In the matched case, exact analytic solution of the

problem proves to be more involved, though some

progress has been made (see the appendix). Never- theless, the form of AEconf is easily deduced.

Again, at large (ip) distances, ð.Econf will be asymp-

totic to

whereas the interaction of the midgap states will

dominate at small separations (see Fig. 4a). AE,:.nf

for the matched case is shown in figure 4b.

Thus in both cases, the solitons experience a confi-

nement energy tending to pin them together. The midgap interaction produces a relatively strong but local confinement energy in the matched case (of depth t.1 ~ 0.1 eV

=

1 150 K, and extent - ± 2.5 ç =

± 35 a), while the confinement energy in the mis- matched case is given by W

=

t2/7rto _ 1.3 x

10- 3 eV = 15 K per site. Even if they are not pinned

to impurities the midgap interaction serves to streng- then the hypothesis that solitons within crystalline regions will be well localized. The above results imply that, at room temperature, they will be localized within - 100 lattice constants in the mismatched case, and within - 10 lattice constants for matched chains.

Although the interaction of the midgap states

strengthens the hypothesis that solitons within (CH)x

are effectively pinned by 3 D effects, it does produce

some other interesting consequences. One such effect

can be illustrated by a simple model : consider 2 (CH)x

chains as in figure 5, one containing a large number of

solitons in fixed positions, and the other one mobile defect. If the concentration of solitons on chain 1 is greater than some critical concentration pc (as is illustrated) the most stable positions for the mobile soliton will be at a locally matched position (i.e. at positions 1, 3, or 5). However, if the concentration is less than pc, the minima at positions 2, 4, and 6 will

have the lowest energy, and local mismatched ordering

will be preferred. Clearly, this argument can be

generalized to two arrays of solitons, and to more complicated systems. Thus, even though mismatched ordering is preferred for pristine chains, matched ordering will be equally relevant in the presence of solitons.

The consequences of the preference of mismatched

ordering have been pursued by Bohr and Brazovskij [12] through a mapping to an Ising model. Their treatment neglects the competing potential wells at

matched order, which would give a strong four spin

term in their model Hamiltonian.

A rough estimate for the critical concentration can

easily be found Assuming that the solitons are equally spaced, a distance I appart, then from figure 5

i.e.

which gives

i.e. roughly one soliton defect per 200 carbon atoms.

Even though in reality the situation will be far

more complicated than this, this does illustrate that the local soliton density will effect the local ordering (i.e. either matched or mismatched), and, therefore,

the size and position of the bandgap.

Fig. 5.

-

Soliton density effects on local chain ordering.

(7)

136

4. Discussion and conclusion.

The model considered [7] is the simplest possible form

of interchain hopping, and we have further restricted the discussion to a pair of chains. Not only the total

energy but also the size and position of the band gap

depends on the matching of the dimerization, as

summarized in table I.

Thus, interchain hopping results in mismatched ordering being preferred by - 15 K per atom. The

bandgap (and its position) varies by 2 t I - 0.2 eV, depending on the local environment of the (CH)x

chains.

The confinement energy of two solitons on parallel (CH)x chains can be studied in the continuum model

of TLM [3]. The exact wavefunctions available in this model depend explicitly on the dimerization pattern

resulting in different results in the matched (same

dimerization pattern at + 00) and the mismatched

cases (opposite dimerisation pattern at + oo). In

both cases, at large separations, Econf is dominated

by the effect of either creating or destroying lengths

of the preferred mismatched AB phase. However,

as the solitons approach each other, the interaction of their localized midgap states becomes important.

The splitting of the midgap states is given by

resulting in a locally large soliton confinement energy in the matched case (see Fig. 5b). This strengthens

the hypothesis that solitons in crystalline (CH)x

will be effectively pinned by 3 D effects (even if they

are not pinned to impurities).

Such splitting of the midgap levels implies that the spin concentration should be partially quenched due

to the pairing of electrons in the lower binding state

of interacting matched solitons (which will always

be present to some extent in any real material). The

characteristic temperature, Tq, however, is of the order

well outside the range of experiments on (CH)x.

At room temperature a soliton will be pinned to

-

100 lattice units (mismatched) and to within - 10

lattice units (matched case). This distinction in confi- nement length will be mirrored by the soliton mobi-

lities (the solitons appearing more mobile in the very shallow wells of the mismatched case), which should be detectable by motional narrowing in ESR measure-

ments (13). The form of Econf also implies the existence of a critical concentration of solitons, pc (~ 1 soliton

per 200 carbon atoms) above which local matched

ordering becomes more favourable than mismatched

ordering. This implies that there should be an upper limit on spin density in a system of neutral solitons,

of approximately 0.05 spins per carbon atom (as

above pc the chains match, the midgap levels split,

and the electrons at midgap become paired). Thus, above pc one would expect a marked decrease in spin density, while the optical absorption should become

« polaron-like » due to the splitting of the midgap

levels.

Unfortunately, however, it is not possible to control

the concentration of neutral solitons in polyacetylene ;

and as the concentration in the pristine Shirakawa

material is approximately 1 in 3 000 [13], well bellow Pc’ testing of the above theory does not seem possible

with present materials.

In conclusion, the splitting of the midgap levels,

in that it predicts even stronger 3D pinning of solitons,

serves to strengthen the case against the solitonic

model for the electronic properties of polyacetylene.

However, it does produce some other interesting

effects on the dependence of local ordering on soliton density, suggesting that matched ordering is just as

relevant as mismatched ordering in the theory of

interchain coupling in polyacetylene.

Acknowledgments.

The present work was supported by a Science and

Engineering Research Council Case Award (Reference

Number 83505170) in conjunction with I.C.I. New Science Group, Runcorn.

Appendix.

ANALYTIC SOLUTION OF SOLITON CONFINEMENT.

-

As mentioned in section 3.3 the mismatched case has

already been solved analytically by Baeriswyl and

Maki [7].

The Hamiltonian for the two interacting chains

is given by

Table I.

-

Band structure for two pristine (CH)x chains.

(8)

where H ff is given by

And so, in the basis 0 = (lkll t/J 2)’ Heff for the two

chain system is given by

so that the perturbation, H’ is given by

As H’ does not connect individual chains to them- selves, first order perturbative effects will be zero.

Thus,

where the factor of 4 comes from the spin summation

over both chains, and the first term comes from intraband transitions, and the second from valence band to midgap transitions.

i.e.

a) Mismatched case.

In this case it is straightforward to show that [7]

which is independent of l, and therefore does not contribute to the confinement energy.

Also, after some manipulation [14] using the boundary conditions given in equations (13),

which, after transforming sums to integrals, and some

further simplifications [7], gives

b) Matched case.

The analysis in this case becomes more complicated.

It is straightforward to evaluate Mk and Mkk, using

the appropriate wavefunctions, giving

and, where

However, further analytic evaluation of 4E from equation (A. 6) appears to be difficult.

References

[1] SU, W. P., SCHRIEFFER, J. R. and HEEGER, A. J., Phys.

Rev. Lett. 42 (1979) 1698, Phys. Rev. B 22 (1980)

2099.

[2] RICE, M. J., Phys. Rev. Lett. 71A (1979) 152.

[3] TAKAYAMA, H., LIN-LIU, Y. R. and MAKI, K., Phys.

Rev. B 21 (1980) 2388.

[4] BAERISWYL, D., Helv. Phys. Acta 56 (1983) 639.

[5] GRANT, P. M. and BATRA, I. P., Solid State Commun.

29 (1979) 225.

[6] FINCHER, C. R., PEEBLES, D. L., HEEGER, A. J., DRUY, M. A., MATSUMURA, Y., MACDIARMID, A. G., SHIRAKAWA, H. and IKEDA, S., Solid State Com-

mun. 27 (1978) 489.

[7] BAERISWYL, D. and MAKI, K., Phys. Rev. B 28 (1983)

2068.

[8] MAKI, K., To Appear in Proc. 2nd International Conf.

Synthetic Metals (Los Alamos) 1983.

[9] JEYADEV, S., Phys. Rev. B 28 (1983) 3447.

[10] BAUGHMAN, R. H. and Moss, G., J. Chem. Phys. 77 (1982) 5321.

[11] GRANT, P. M. and BATRA, I. P., J. Physique Colloq.

44 (1983) C3-437.

[12] BOHR, T. and BRAZOVSKIJ, S. A., J. Phys. C. 16 (1983)

1189.

[13] GOLDBERG, I. B., CROWE, H. R., NEWMAN, P. R., HEEGER, A. J. and MACDIARMID, A. G., J. Chem.

Phys. 70 (1979) 1132.

[14] KIVELSON, S., TING-KUO LEE, LIN-LIU, Y. R., PESCHEL,

I. and LU YU, Phys. Rev. B. 25 (1982) 4173.

Références

Documents relatifs

This study summarizes the refit of the 8.2 m LOA Government of Nunavut owned research vessel RV Papiruq to make it suitable for fisheries research in coastal (i.e.,

intersoliton hopping /lo/: there are charged and neutral solitons, both free to move along their chain segments, but they cannot leave these segments.. They can, however,

Solutions of the dynamical equations that are obtained in a semidiscrete approximation, have the form of circularly polarized and linearly polarized envelope modes..

On the origin of a strong central peak in the local density of d states at the surface of transition

<v 0.5) observed in some vitreous systems/l ,2/ cannot be interpreted by a constant density of two level defects /3,4/ leading to a linear temperature depen- dence. Here we

Concerning the Cauchy problem for the hydrodynamical Landau-Lifshitz equation, the main difficulty is to establish the continuity with respect to the initial datum in the energy space

The scaling coefficients of the higher order time variables are explicitly computed in terms of the physical parame- ters, showing that the KdV asymptotic is valid only when the

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des